BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on September 27, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Sleep problems and related influencing factors among healthcare workers in tertiary public hospitals in Heilongjiang province ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-032239

Article Type: Research

Date Submitted by the 09-Jun-2019 Author:

Complete List of Authors: liu, huan; Harbin Medical University, Liu, jing; Harbin Medical University School of Public Health, Department of Social Medicine chen, mingxi; Harbin Medical University, Social Medicine tan, xiao; Harbin Medical University, Social Medicine zheng, tong; Harbin Medical University, Social Medicine Kang, Zheng; Harbin Medical University, Social Medicine; Gao, Lijun; Harbin Medical University, Social Medicine Jiao, Mingli; Harbin Medical University, Health policy Ning, Ning; Harbin Medical University, Social Medicine Liang, Libo; School of Health Management, Harbin Medical University, Social Medicine; Wu, Qunhong; Harbin Medical University, Social Medicine http://bmjopen.bmj.com/ Hao, Yanhua; Harbin Medical University, Social Medicine

Healthcare worker, Tertiary hospitals, Sleep problems, Influencing Keywords: factors

on September 27, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Title Page 5 6 7 2 8 9 3 Title 10 11 4 Sleep problems and related influencing factors among healthcare workers in tertiary 12 5 public hospitals in Heilongjiang province. 13 14 6 15 16 17 7 Author names, affiliations and email addresses 18 For peer review only 19 8 Huan Liu1, Jingjing Liu1, Mingxi Chen1, Xiao Tan1, Tong Zheng1, Zheng Kang1, 20 9 Lijun Gao1, Mingli Jiao2, Ning Ning1, Libo Liang1, Qunhong Wu1*, Yanhua Hao1*. 21 22 10 23 24 25 11 1Department of Social Medicine, School of Public Health, Harbin Medical University, 26 27 12 Harbin, Heilongjiang Province, China. 28 29 30 13 2Department of Health Policy, School of Public Health, Harbin Medical University, 31 32 33 14 Harbin, Heilongjiang Province, China. 34 35 15 36

37 16 http://bmjopen.bmj.com/ 38 39 17 Huan Liu, Department of Social Medicine, School of Public Health, Harbin Medical 40 41 18 University, Harbin, Heilongjiang Province, China. Email: [email protected] 42 43 19 Jingjing Liu, Department of Social Medicine, School of Public Health, Harbin Medical 44 20 University, Harbin, Heilongjiang Province, China. Email: [email protected]

45 on September 27, 2021 by guest. Protected copyright. 46 47 21 Mingxi Chen, Department of Social Medicine, School of Public Health, Harbin 48 49 22 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 50 Xiao Tan, Department of Social Medicine, School of Public Health, Harbin Medical 51 23 52 24 University, Harbin, Heilongjiang Province, China. Email: [email protected] 53 54 25 Tong Zheng, Department of Social Medicine, School of Public Health, Harbin Medical 55 56 26 University, Harbin, Heilongjiang Province, China. Email: 57 58 27 [email protected] 59 60 28 Zheng Kang, Department of Social Medicine, School of Public Health, Harbin Medical

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 University, Harbin, Heilongjiang Province, China. Email: [email protected] 5 6 2 Lijun Gao, Department of Social Medicine, School of Public Health, Harbin Medical 7 8 3 University, Harbin, Heilongjiang Province, China. Email: [email protected] 9 10 4 Mingli Jiao, Department of Health Policy, School of Public Health, Harbin Medical 11 University, Harbin, Heilongjiang Province, China. Email: [email protected] 12 5 13 6 Ning Ning, Department of Social Medicine, School of Public Health, Harbin Medical 14 15 7 University, Harbin, Heilongjiang Province, China. Email: [email protected] 16 17 8 Libo Liang, Department of Social Medicine, School of Public Health, Harbin Medical 18 For peer review only 19 9 University, Harbin, Heilongjiang Province, China. Email: [email protected] 20 21 10 Qunhong Wu, Department of Social Medicine, School of Public Health, Harbin 22 23 11 Medical University, Harbin, Heilongjiang Province, China. Email: 24 25 12 [email protected] 26 27 13 Yanhua Hao, Department of Social Medicine, School of Public Health, Harbin 28 29 14 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 30 31 15 32 33 16 *Corresponding authors: 34 35 17 Qunhong Wu 36 18 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District,

37 http://bmjopen.bmj.com/ 38 39 19 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 40 41 20 87502851; fax: + 86-0451-87502853 42 43 21 44 22 Yanhua Hao

45 on September 27, 2021 by guest. Protected copyright. 46 47 23 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District, 48 49 24 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 50 87502860; fax: + 86-0451-87502853 51 25 52 26 53 54 55 27 Keywords: Healthcare worker; Tertiary hospitals; Sleep problems; Influencing factors 56 57 28 Word counts 58 59 29 word count of the abstract: 276 60 30 word count of manuscript: 3934

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Sleep problems and related influencing factors among healthcare 5 6 7 2 workers in tertiary public hospitals in Heilongjiang province: a 8 9 3 multi-level analysis 10 11 12 4 13 14 15 5 Abstract 16 17 6 Objectives: Sleep problems not only affect the health and work quality of healthcare 18 For peer review only 19 20 7 workers, but also it might bring risks to patients’ safety. China's tertiary hospitals 21 22 8 provide medical care for a large proportion of the Chinese population and therefore, 23 24 25 9 sleep quality of healthcare workers in these institutions increasingly becomes an issue 26 27 10 of paramount concern. We aim to conduct subjective evaluation on the sleep problems 28 29 30 11 and explore potential influencing factors among healthcare workers in tertiary hospitals 31 32 33 12 in China. 34 35 13 Design: A cross-sectional questionnaire survey study. 36

37 http://bmjopen.bmj.com/ 38 14 Setting: The study was conducted in 377 departments at 33 tertiary hospitals in 39 40 15 Heilongjiang Province. 41 42 43 16 Participants: A total of 4009 participants were invited. Ultimately, 3,810 participants 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 completed valid questionnaires. Study participants included clinicians, medical 47 48 18 technicians, and administrative staffs. 49 50 51 19 Primary and secondary outcome measures: Participants completed a structured 52 53 20 questionnaire, which collected data regarding demographics, sleep problems, 54 55 56 21 employment, chronic pain, and healthy behaviors. A two-level logistic regression 57 58 59 22 model was conducted to examine determinants of sleep problems. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Results: In this study, we found that nearly half of the healthcare workers reported 5 6 7 2 having sleep problems. The two-level logistic model suggested a positive association 8 9 3 of sleep problems with headache (OR = 2.61, 95%CI = 2.24–3.04), working hours (OR 10 11 12 4 = 1.39, 95%CI = 1.18–1.64), and night shifts (OR=1.44, 95%CI=1.22-1.71). In contrast, 13 14 5 regular diet (OR = 0.50, 95%CI = 0.41–0.62), regular bowel movements (OR = 0.70, 15 16 17 6 95%CI = 0.56–0.87), and exercise (OR = 0.66, 95%CI = 0.54–0.81) protected against 18 For peer review only 19 20 7 sleep problems in healthcare workers. 21 22 8 Conclusions: Sleep problems are prominent among healthcare workers. The leading 23 24 25 9 factor associated with sleep problems was headache. Future studies should explore the 26 27 10 effects of interventions for sleep problems among healthcare workers. 28 29 30 11 Strengths and limitations of this study 31 32 33 12 · We recruited 3,810 healthcare workers from 377 departments at 33 tertiary hospitals 34 35 13 in Heilongjiang Province with a multi-stage stratified random sampling. 36

37 http://bmjopen.bmj.com/ 38 14 · Use a two-level logistic regression model to examine determinants of sleep problems. 39 40 15 · Information regarding the exposure and outcome was obtained via a self-report 41 42 43 16 questionnaire, which may introduce report bias in these variables. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 · The results of the study are not generalizable to all Chinese doctors. 47 48 18 · The cross-sectional design limited our ability to establish causal relationships based 49 50 51 19 on the results. 52 53 20 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Introduction 5 6 7 2 Sleep problems has become an important public health problem. Globally, it is 8 9 3 estimated that 7.6% of the global population suffered from sleep problems 1. In China, 10 11 12 4 many occupations are at risk of sleep problems and sleep problems among healthcare 13 14 5 workers were particularly evident 2. Many scholars have reported a high prevalence of 15 16 17 6 sleep problems among Chinese healthcare workers, ranging from 21% to 65.5% 7-9. 18 For peer review only 19 20 7 Some previous studies have shown that sleep problems are common among healthcare 21 22 8 workers including clinicians, nurses, medical technicians, and administrative staff 23 24 25 9 worldwide. 26 27 10 For example, Hayashino et al. observed that 30.7% of Japanese healthcare workers 28 29 30 11 reported often having difficulty in initiating sleep 3. In the United States, Patterson et 31 32 33 12 al. reported that 59.2% of members of the National Emergency Medical Service 34 35 13 Management Association had poor sleep quality 4. Moreover, a research in Italy found 36

37 http://bmjopen.bmj.com/ 38 14 that the prevalence of sleep disturbances among healthcare workers was extremely high, 39 40 15 even among healthy controls in the study, with 24.6% reporting difficulties in falling 41 42 43 16 asleep and 31.7% experiencing frequent awakenings during the night 5. A study in Iran 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 observed that almost 43% of healthcare workers reported different degrees of sleep 47 48 18 problems 6. 49 50 51 19 It is crucial to pay special attention to sleep problems among healthcare workers, the 52 53 20 main reasons are as follows: First, sleep problems can affect healthcare workers’ health 54 55 56 21 and safety. For example, many previous studies have confirmed that sleep problems 57 58 59 22 have a powerful influence on the risk of cardiovascular disease, immune disorders, and 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 cancer 7-9. Sun and colleagues have reported a significant negative correlation between 5 6 10 7 2 anxiety and sleep problems in healthcare workers . Moreover, Patterson et al. reported 8 9 3 that injuries occurred approximately 2 times more frequently among emergency 10 11 12 4 medical workers who have poor sleep quality 11. Second, sleep problems not only affect 13 14 5 the health of healthcare workers themselves, but also might affect patient safety and 15 16 17 6 quality of medical services. Indeed, Eddy showed clear evidence indicating that sleep 18 For peer review only 19 20 7 deprivation caused by long work hours can affect both the health of healthcare workers 21 22 8 and the safety of patients 12. Some previous studies have found that sleep deprivation, 23 24 25 9 long work hours, and night shift can lead to medical errors such as an increase in the 26 27 10 rate of surgical complications, needlestick injuries, adverse drug events, and diagnostic 28 29 30 11 errors 13-16. 31 32 17 18 33 12 The acceleration of the aging process and the increase in the incidence of cancer 34 35 13 and cardiovascular diseases 19 have put tremendous pressure on the health care system 36

37 http://bmjopen.bmj.com/ 38 14 in China. In response to increasing demands on the health services, healthcare workers 39 40 15 have a bigger workload and extended working hours. The previous study have indicated 41 42 43 16 that long working hours were the critical determinant of sleep health 20. However, few 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 studies have focused on the relationship between work hours and sleep problems in 47 48 18 healthcare workers in China. Alemohammad et al. reported that shift healthcare workers 49 50 51 19 have poorer sleep quality than non-shift healthcare workers and highlighted that poor 52 53 20 sleep quality was aggravated by shift work 21. Hence, frequent night shifts are likely to 54 55 56 21 be an important factor affecting sleep problems. Furthermore, chronic pain is common 57 58 22 23 59 22 among healthcare workers and has been shown to be related to sleep problems . 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Nevertheless, few studies have investigated which specific site of chronic pain has the 5 6 7 2 greatest impact on sleep problems. Healthy behaviors are also an import potential 8 9 3 influencing factor. For example, some research has shown that unhealthy diets are 10 11 12 4 associated with sleep problems and healthy diet can effectively improve sleep quality 13 14 5 24. Furthermore, a recent study conducted in Japan observed that healthcare workers 15 16 17 6 who reported often having trouble initiating sleep consumed more alcohol and 18 For peer review only 19 3 20 7 exercised less . 21 22 8 In 2009, China launched a nationwide systematic reform of healthcare services. 23 24 25 9 Although the reform had been successful in some areas, much still need to be done to 26 27 10 reform China’s healthcare delivery. Particularly, primary healthcare facilities have not 28 29 30 11 been able to perform a gate-keeping function and most patients in China prefer to visit 31 32 33 12 tertiary hospitals, especially tertiary public hospitals, even for common and minor 34 35 13 illnesses. In 2016, there were about 3.27 billion patient visits to Chinese hospitals, 49.8% 36

37 http://bmjopen.bmj.com/ 38 14 of which were visits to tertiary hospitals 25. The results of a survey of the Chinese 39 40 15 Medical Doctor Association in 2014 showed that 92% of healthcare workers in tertiary 41 42 43 16 hospitals had to work overtime, and 72% of healthcare workers who have worked more 44

45 26 on September 27, 2021 by guest. Protected copyright. 46 17 than 60 hours a week on average are in tertiary hospitals in China . Therefore, sleep 47 48 18 quality of healthcare workers in tertiary hospitals is of paramount importance. 49 50 51 19 Nevertheless, population-based prevalence data on sleep problems among healthcare 52 53 20 workers in tertiary hospitals in China are limited. 54 55 56 21 In summary, our study has the following advantages. First, although people have 57 58 59 22 realized that sleep problems in healthcare workers are common, studies on healthcare 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 workers in tertiary hospitals in China are limited. Second, studies explore only some 5 6 7 2 factors that may influence sleep problems of healthcare workers. However, the 8 9 3 relationship between sleep problems and work factors, healthy behavior, and chronic 10 11 12 4 pain remains to be discovered. Third, compared to previous studies 21 27-29, our study is 13 14 5 a multi-center study that covers a large sample of healthcare workers in the 15 16 17 6 Heilongjiang province, China. 18 For peer review only 19 20 7 The purpose of the present study was to investigate the prevalence of self-reported sleep 21 22 8 problems among healthcare workers in tertiary public hospitals, so as to identify 23 24 25 9 potential factors influencing sleep problems among healthcare workers in tertiary 26 27 10 public hospitals. 28 29 11 30 31 12 Methods 32 33 34 13 Data collection and study population 35 36 14 This study was undertaken in Heilongjiang province, in northeast China, where there

37 http://bmjopen.bmj.com/ 38 39 15 were 689 public hospitals in 2014, of which 82 were tertiary public hospitals 30. This 40 41 42 16 study focused on healthcare workers in representative tertiary hospitals. The study 43 44 17 participants included clinicians (physicians, surgeons, pediatricians, gynecologists,

45 on September 27, 2021 by guest. Protected copyright. 46 47 18 emergency doctor, etc.), medical technicians (pharmacists, radiological technicians, 48 49 50 19 laboratory physicians, etc.), and office workers (administrative department personnel, 51 52 20 finance personnel, etc.). First, 33 tertiary public hospitals were selected at random to 53 54 55 21 serve as survey locations for this study. For the purposes of reporting, hospitals are not 56 57 22 identified by name. Second, the departments of each hospital were coded and several 58 59 60 23 departments were randomly selected from each hospital. The sample size at each level

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 was based on the proportion of the number of workers in the hospital at that level. In 5 6 7 2 the final sample, 4,007 healthcare workers from 317 departments were invited to 8 9 3 participate in the study. A self-administered questionnaire was completed by healthcare 10 11 12 4 workers in the selected departments (Additional file 1). After the data cleaning process, 13 14 5 incomplete or anomalous data (n = 197) were excluded, leaving a final sample of 3,810 15 16 17 6 participants. 18 For peer review only 19 20 7 21 22 8 Variables 23 24 25 9 A structured questionnaire was designed to obtain data regarding demographics, sleep 26 27 10 problems, working conditions, chronic pain, and healthy behaviors. 28 29 30 11 31 32 33 12 Outcome measure (dependent variable) 34 35 13 The main outcome was the presence of sleep problems, which was a binary variable (1 36

37 http://bmjopen.bmj.com/ 38 14 = yes; 0 = no). The presence of sleep problems was assessed from the participant’s self- 39 40 15 report. Participants were asked the question “Overall in the last 30 days, how much of 41 42 43 16 a problem did you have with sleeping, such as falling asleep, waking up frequently 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 during the night, or waking up too early in the morning?” with the following five answer 47 48 18 options: none, mild, moderate, severe, and extreme. Participants who answered extreme 49 50 51 19 and severe were considered to have sleep problems. This definition has been used in 52 53 20 previous publications using the same survey question on sleep problems 1 31. 54 55 56 21 57 58 59 22 Individual-level variables (independent variables) 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Individual-level variables included demographic indicators (age, sex, educational 5 6 7 2 background, job title, and monthly salary income), working conditions (average weekly 8 9 3 working hours, number of night shifts per month, and one-way commute time), healthy 10 11 12 4 behaviors (physical exercise, and regularity of diet and bowel movements during work), 13 14 5 and experience of chronic pain (headache, neck and shoulder pain, lumber and back 15 16 17 6 pain, and knee pain). 18 For peer review only 19 20 7 We coded sex as women (= 0) and men (= 1). Meanwhile, educational background was 21 22 8 coded as bachelor’s degree and below (= 0) and master’s degree and above (= 1). Age 23 24 25 9 was categorized into three age groups (< 30, 30–40, > 40 years) and job title was also 26 27 10 an ordinal categorical variable (junior titles and others, intermediate job title, and senior 28 29 30 11 title). Therefore, we needed to judge whether the two variables are introduced into the 31 32 33 12 model as dummy variables or grouped linear variables according to the fit of model. 34 35 13 The median monthly salary income of participants in this study was 3,000 yuan. 36

37 http://bmjopen.bmj.com/ 38 14 Therefore, we coded monthly salary income as less or equal to 3,000 yuan (= 0) and 39 40 15 above 3,000 yuan (=1). 41 42 43 16 One-way commute time was a dummy variable, coded as less than 30 minutes (= 0) 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 and as 30 minutes and above (= 1). The median of average weekly working hours of 47 48 18 participants in this study was 55 hours. Therefore, average weekly working hours were 49 50 51 19 evaluated with the question: “How much time do you work on average per week?” The 52 53 20 response categories were 1 = 55 hours and above, and 0 = less than 55 hours. Whether 54 55 56 21 participants have completed night shifts was measured with the question: “What's your 57 58 59 22 average monthly night shift?” The response categories were 1 = 5 times or less, and 0 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 = more than 5 times. 5 6 7 2 Regularity of meals was assessed with the question, “Was your diet regular at work?” 8 9 3 The response options were 1 = yes and 0 = no. Regularity of bowel movements was 10 11 12 4 assessed with the question, “Did you defecate regularly at work?” The response options 13 14 5 were 1 = yes and 0 = no. Physical exercise was assessed with the question “Do you 15 16 17 6 engage in physical exercise every day?” where 1 = yes and 0 = no. 18 For peer review only 19 20 7 We also asked participants whether or not they have chronic pain in any of the following 21 22 8 regions in the last 30 days: headache, neck and shoulder, lumbar, or knee. The response 23 24 25 9 categories were 1 = yes and 0 = no. 26 27 10 28 29 30 11 Department-level variables (independent variable) 31 32 33 12 The participant’s department type was a level-2 explanatory variable in our analysis 34 35 13 because individuals in the same department are not independent of each other. Among 36

37 http://bmjopen.bmj.com/ 38 14 the healthcare workers in the study, some practitioners were from clinical departments 39 40 15 (including internal medicine, surgery, gynecology, pediatrics, etc.) and others were 41 42 43 16 health technicians (including laboratory, pathology, radiology, etc.). Moreover, some 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 health personnel from administrative departments also participated in the survey. 47 48 18 Therefore, this study is based on the functional attributes of the department, and 49 50 51 19 whether the participants are from a clinical department (1 = yes, 0 = no) was an 52 53 20 explanatory variable. 54 55 56 21 57 58 59 22 Data analysis 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 In this study, we used two statistical methods. The 2-test was used to examine the 5 6 7 2 independent associations between participants’ sleep problems and each independent 8 9 3 variable. Since the data were hierarchical, participants (level-1) were nested within 10 11 12 4 departments (level-2), to create a two-level logistic model that was applied to assess 13 14 5 associations between sleep problems and study variables. 15 16 17 6 Intra-class correlation coefficients (ICC) for healthcare workers were calculated to 18 For peer review only 19 32 20 7 assess clustering by department . First, an empty model was developed to verify 21 22 8 whether our research model is applicable to the two-level logistic model. If the ICC was 23 24 25 9 significant, then it suggested that a multi-level logistic regression model was 26 27 10 appropriate. Subsequently, Model 1 added the department-level explanatory variable to 28 29 30 11 the empty model. Next, Model 2 added individual-level explanatory variables to Model 31 32 33 12 1. We compared the fit of models with the difference in −2 log-likelihood (or deviance 34 35 13 statistics) between the two models. A statistically significant difference in −2 log- 36

37 http://bmjopen.bmj.com/ 38 14 likelihood between the two models indicated that the fit of the latter model was better 39 40 15 33. 41 42 43 16 Two-level binary logistic regression models were estimated with the GLIMMIXED and 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 NLMIXED procedures of SAS 9.4 (SAS Institute Inc., Cary, NC, USA). All the other 47 48 18 analyses were conducted in SPSS 21.0 (IBM Corp, Armonk, USA), and values of p < 49 50 51 19 0.05 were considered significant. 52 53 20 54 55 56 21 Ethics approval 57 58 59 22 The study protocol was reviewed and approved by the Research Ethics Committee of 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Harbin Medical University and the ethics committees from each hospital. The study 5 6 7 2 also received permission and assistance from each hospital to contact their healthcare 8 9 3 workers. All participants were informed that the study was voluntary and confidential. 10 11 12 4 Participants were informed that they have the right to discontinue their participation 13 14 5 or withdraw their consent at any time and were not obliged to state their reasons. 15 16 17 6 After provided written informed consent by participants, interviews were conducted. 18 For peer review only 19 20 7 Survey data were collected using an anonymous questionnaire to protect the privacy 21 22 8 of participants. 23 24 25 9 26 27 10 Patient and public involvement 28 29 30 11 No patients were involved. 31 32 33 12 34 35 13 Results 36

37 http://bmjopen.bmj.com/ 38 14 Participants’ characteristics 39 40 15 Participants’ general characteristics are shown in Table 1. Overall, participants are 41 42 43 16 generally unable to maintain a regular diet (75.5%) and regular bowel movements at 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 work (80.4%). The most common types of chronic pain were neck and shoulder pain 47 48 18 (69.9%) and lumber and back pain (64.1%). Sleep problems were reported by 48.5% of 49 50 51 19 participants. The univariate analysis indicated that demographics (excluding sex and 52 53 20 monthly salary income), chronic pain, and working conditions were all related to 54 55 56 21 participants’ sleep problems (p < 0.05). 57 58 59 22 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 1 Table 1. Characteristics of questionnaire respondents and results of the univariate 4 5 2 analysis 6 Variable Total Sleep problems p-value 7 8 N = 3810 (%) No Yes 9 n =1964 (%) n = 1846 (%) 10 11 Department-level 12 Clinical department .001 13 Yes 3,297 (86.5) 1,658 (50.2) 1,639 (49.7) 14 15 No 513 (13.5) 306 (59.6) 207 (40.4) 16 Individual-level 17 Age (years) < .001 18 For peer review only 19 < 30 504 (13.2) 308 (61.1) 196 (38.9) 20 30–40 2,054 (53.9) 982 (47.8) 1,072 (52.2) 21 > 40 1,252 (32.9) 674 (53.8) 578 (46.2) 22 Sex .279 23 24 Men 2,455 (64.4) 1,282 (52.2) 1,173 (47.8) 25 Women 1,355 (35.6) 682 (50.3) 673 (49.7) 26 27 Educational background .004 28 Bachelor degree and below 1,883 (49.4) 926 (49.2) 957 (50.8) 29 30 Master's degree and above 1,927 (50.6) 1,038 (53.9) 889 (46.1) 31 Job title .001 32 33 Junior titles and others 924 (24.3) 521 (48.8) 403 (51.2) 34 Intermediate title 1,287 (33.8) 622 (30.8) 665 (69.2) 35 Senior title 1,599 (42.0) 821 (51.3) 778 (48.7) 36 Monthly salary income .791

37 http://bmjopen.bmj.com/ 38 ≤ 3,000 yuan 2,286 (60.0) 1,112 (48.6) 1,174 (51.4) 39 > 3,000 yuan 1,524 (40.0) 790 (51.8) 734 (48.2) 40 One-way commute time .001 41 42 < 30 minutes 2,038(53.5) 1097 (53.8) 941 (46.2) 43 ≥ 30 minutes 1,772(46.5) 867 (48.9) 905 (51.1) 44 Regular diet at work < .001

45 on September 27, 2021 by guest. Protected copyright. 46 Yes 934 (24.5) 682 (73.0) 252 (27.0) 47 No 2,876 (75.5) 1,282 (44.6) 1,594 (55.4) 48 49 Regular bowel movement at work < .001 50 Yes 747 (19.6) 537 (66.2) 210 (33.8) 51 52 No 3,063 (80.4) 1,427 (34.2) 1,636 (65.8) 53 Work hours per week (hours) < .001 54 ≥ 55 1,964 (51.5) 850 (42.3) 1,114 (56.7) 55 56 < 55 1,846 (48.5) 1,114 (60.3) 732 (39.7) 57 Night shifts per month (times) < .001 58 59 ≤ 5 2,171 (57.0) 1,270 (58.5) 901 (41.5) 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 34 BMJ Open

1 2 3 > 5 1,639 (43.0) 694 (42.3) 945 (57.7) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Daily exercise < .001 6 Yes 685 (18.0) 452 (66.0) 233 (34.0) 7 8 No 3,125 (82.0) 1,512 (48.4) 1,613 (51.6) 9 Headaches < .001 10 11 Yes 1,698 (44.6) 601 (35.4) 1,097 (64.6) 12 No 2,112 (55.4) 1,363 (64.5) 749 (35.5) 13 Neck and shoulder pain < .001 14 15 Yes 2,664 (69.9) 1,243 (46.7) 1,421 (53.3) 16 No 1,146 (30.1) 721 (62.9) 425 (37.1) 17 18 Lumber and back painFor peer review only < .001 19 Yes 2,441 (64.1) 1,111 (45.5) 1,330 (54.5) 20 21 No 1,369 (35.9) 853 (62.3) 516 (37.7) 22 Knee pain < .001 23 Yes 975 (25.6) 340 (34.9) 635 (65.1) 24 25 No 2,835 (74.4) 1,624 (57.3) 1,211 (42.7) 26 1 27 28 2 Self-reported sleep duration 29 30 31 3 Figure 1 shows that the largest proportion of respondents reported sleeping for about 32 33 34 4 6 hours, regardless of department. The larger proportion of healthcare workers who 35 36 5 reported sleep less than or equal to 5 hours, were working in clinical departments.

37 http://bmjopen.bmj.com/ 38 39 6 Moreover, 46.7% of participants reported a sleep duration about 6 hours per day and 40 41 7 11.3% need more sleep. 42 43 44 8

45 on September 27, 2021 by guest. Protected copyright. 46 47 9 Multi-level analysis 48 49 10 Table 2 presents results from the multi-level analyses. The results of running an 50 51 52 11 empty model showed significant between-group variation (ICC = 0.1181, p < 0.001). 53 54 12 An ICC of 0.1181 indicated that about 12% of the total variation in the outcome 55 56 57 13 variables was caused by variation between the departments. 58 59 60 14 Subsequently, the department-level explanatory variable was introduced. The results

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 showed that the odds of having sleep problems was about 1.5 times greater among 5 6 7 2 healthcare workers in clinical departments than among those from non-clinical 8 9 3 departments (OR = 1.53, 95%CI = 1.15–2.04). After we incorporated the department- 10 11 12 4 level explanatory variable, -2 Log Likelihood declined from 5159.4 to 5150.9. The 13 14 5 difference in -2 Log Likelihood between model 1 and the empty model indicated that 15 16 17 6 the fit of model 1 was better (LR = 8.5, df = 1, p < 0.01). 18 For peer review only 19 20 7 In model 2, individual-level variables were added. The results showed that sleep 21 22 8 problems for healthcare workers had a protective association with regular diet (OR = 23 24 25 9 0.50, 95%CI = 0.41–0.62), regular bowel movements (OR = 0.70, 95%CI = 0.56– 26 27 10 0.87), and exercise (OR = 0.66, 95%CI = 0.54–0.81). Compared with participants 28 29 30 11 with a senior job title, participants with junior titles had lower odds of suffering from 31 32 33 12 sleep problems (OR = 0.72, 95%CI = 0.56–0.93). Model 2 further suggested that 34 35 13 chronic pain was negatively associated with sleep problems. Specifically, the results 36

37 http://bmjopen.bmj.com/ 38 14 suggest that headaches are more likely to be reported by participants with sleep 39 40 15 problems (OR = 2.61, 95%CI = 2.24–3.04). Furthermore, working hours (OR = 1.39, 41 42 43 16 95%CI = 1.18–1.64) and night shifts (OR = 1.44, 95%CI = 1.22–1.71) were also 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 associated with sleep problems. The difference in -2 Log Likelihood between model 2 47 48 18 and model 1 indicated that the fit of Model 2 was better (LR = 623, df = 15, p < 49 50 51 19 0.001). Meanwhile, department-level variables were also tested but showed no 52 53 20 statistical significance. 54 55 21 56 57 22 Table 2. Factors related to sleep problems: results from two-level logistic regression 58 Variable Empty model Model 1 Model 2 59 Department-level 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 34 BMJ Open

1 2 3 Clinical department BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 ** 5 Yes 1.53 (1.15, 2.04) 1.07 (0.81, 1.40) 6 No 1 (reference) 1 (reference) 7 Individual-level 8 9 Age (years) 10 < 30 0.84 (0.61, 1.16) 11 30–40 1.04 (0.84, 1.28) 12 13 > 40 1 (reference) 14 Educational background 15 Bachelor degree and below 0.87 (0.73, 1.02) 16 17 Master's degree and above 1 (reference) 18 Job title For peer review only 19 Junior title or no title 0.72 (0.56, 0.93) * 20 21 Intermediate title 0.87 (0.71, 1.08) 22 Senior title 1 (reference) 23 One-way commute time 24 <30 minutes 1.06 (0.91, 1.23) 25 26 ≥30 minutes 1 (reference) 27 Regular diet at work 28 Yes 0.50 (0.41, 0.62) *** 29 30 No 1 (reference) 31 Regular bowel movements at work 32 Yes 0.70 (0.56, 0.87) ** 33 34 No 1 (reference) 35 Daily exercise 36 Yes 0.66 (0.54, 0.81) ***

37 No 1 (reference) http://bmjopen.bmj.com/ 38 39 Work hours per week 40 ≥ 55 1.39 (1.18, 1.64) *** 41 < 55 1 (reference) 42 43 Night shifts per month 44 ≤ 5 1.44 (1.22, 1.71) ***

45 > 5 1 (reference) on September 27, 2021 by guest. Protected copyright. 46 47 Headache 48 Yes 2.61 (2.24, 3.04) *** 49 No 1 (reference) 50 Neck and shoulder pain 51 ** 52 Yes 1.27 (1.08, 1.50) 53 No 1 (reference) 54 55 Lumbar and back pain 56 Yes 1.34 (1.14, 1.57) *** 57 No 1 (reference) 58 59 Knee pain 60 Yes 1.60 (1.33, 1.91) ***

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 34

1 2 3 No 1 (reference) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Sigma μ (SE) 0.4407 (0.092) 0.4061 (0.088) 0.2185 (0.067) 6 Intra-class correlation 0.1181*** 0.1099*** 0.0623*** 7 −2 log-likelihood 5159.4 5150.9 4527.9 8 * ** *** 9 1 p < 0.05; p < 0.01; p < 0.001 10 2 11 12 3 Influencing factors of headaches 13 14 15 4 Table 3 shows influencing factors of headaches among respondents with sleep problems. 16 17 5 We found that 62.4% of participants with headache worked 55 hours or more per week, 18 For peer review only 19 20 6 73.5% of respondents with headache do not exercise every day. 21 22 7 23 8 Table 3. Influencing factors of headache among respondents with sleep problems 24 9 (n=1846) 25 26 Variable Headache p-value 27 Yes No 28 n= 1097 (%) n= 749 (%) 29 30 Work hours per week .015 31 ≥55 hours 685(62.4) 429(57.3) 32 33 <55 hours 412(37.6) 320(42.7) 34 Number of night shifts per .173 35 month 36 5 times or less 525(47.9) 376(50.2)

37 http://bmjopen.bmj.com/ 38 More than 5 times 572(52.1) 373(49.8) 39 40 Do exercise every day < .001 41 Yes 120 (10.9) 113(15.1) 42 No 977 (89.1) 636(84.9) 43 44 Regular diet at work < .001

45 Yes 122 (11.12) 130 (17.36) on September 27, 2021 by guest. Protected copyright. 46 47 No 975 (88.88) 619 (82.64) 48 Neck and shoulder pain < .001 49 50 Yes 880 (80.2) 540 (72.1) 51 No 217 (19.8) 209 (27.9) 52 Lumbar and back pain < .001 53 54 Yes 833 (75.9) 496 (66.2) 55 No 264 (24.1) 253 (33.8) 56 57 Knee pain < .001 58 Yes 498 (45.4) 136 (18.2) 59 No 559 (54.6) 613 (81.8) 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Discussion 5 6 7 2 This study contributes to current knowledge of the seriousness of sleep problems for 8 9 3 healthcare workers by reporting the prevalence of sleep problems among healthcare 10 11 12 4 workers in tertiary public hospitals in Heilongjiang province and identifying factors 13 14 5 that can influence sleep problems of healthcare workers. Our results show that more 15 16 17 6 than 45% of healthcare workers reported sleep problems, much higher than in the 18 For peer review only 19 2 20 7 general population of China . 21 22 8 The leading negative factor associated with sleep problems among healthcare workers 23 24 25 9 was headache. In this study, we found that 44.6% of participants experienced 26 27 10 headache in the past month. Previous studies have also found that headache is a very 28 29 30 11 common symptom among healthcare workers. Sokolovic et al. found that 61% of 31 32 33 12 healthcare workers have experienced at least one type of headache in the previous 34 35 13 three months. A cross-sectional study of healthcare workers conducted in Enugu, 36

37 http://bmjopen.bmj.com/ 38 14 South East Nigeria showed that 12% of participants had severe headache that affected 39 40 15 their sleep. 34. Some scholars have suggested from a clinical perspective that the 41 42 43 16 relationship between headaches and sleep problems may be bidirectional 35. The 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 treatment of headache among medical personnel is crucial to relieve sleep problems, 47 48 18 and the improvement of sleep is likely to also reduce headache symptoms of medical 49 50 51 19 personnel. In addition, we analyzed the possible causes of the participants’ headaches 52 53 20 and found that among respondents with sleep problems, factors of irregular diet, work 54 55 56 21 overtime, lack of exercise, neck and shoulder pain, lumbar and back pain and knee 57 58 59 22 pain were associated with headache. Therefore, we need to pay more attention to 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 address factors and to relieve the common problems of headaches of participants. 5 6 7 2 Furthermore, our study confirms the negative effects of other types of chronic pain on 8 9 3 sleep, suggesting that chronic pain in different parts of the body increases the 10 11 12 4 likelihood of sleep problems. Andersen et al. in research among healthcare workers, 13 14 5 have also indicated that, chronic musculoskeletal pain in one body region increases 15 16 17 6 the risk of developing chronic pain in other pain-free body regions 22. Therefore, 18 For peer review only 19 20 7 when healthcare workers complain of chronic pain, they should be diagnosed and 21 22 8 treated in time, which not only reduces the risk of sleeping problems, but also avoids 23 24 25 9 causing chronic pain in other parts. 26 27 10 Our finding that night-shift work predicted a greater risk of sleep problems is 28 29 30 11 consistent with several previous studies 6 21. In hospital, inpatient services usually 31 32 33 12 require round-the -clock services, which often means healthcare workers have to work 34 35 13 shifts (especially night shifts) and long hours, which deprive them of sleep. Moreover, 36

37 http://bmjopen.bmj.com/ 38 14 night work leads to misalignments between internal circadian rhythms and work-rest 39 40 15 schedules, which again leads to fewer sleeping hours 36. Find effective ways to 41 42 43 16 minimize the negative impact of shift work on health workers sleep is an important 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 task. Previous research suggests that timed bright lights and light blocking goggles 47 48 18 can promote circadian adaption to shift work 37 38. However, more reasonable night 49 50 51 19 work schedules for healthcare workers may be a better solution to improve their sleep 52 53 20 problems. 54 55 56 21 We also observed that long working hours were a critical risk factor associated with 57 58 59 22 sleep problems. A cross-sectional study in the United States reported that 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 compensated working time was consistently associated with sleep duration, such that 5 6 39 7 2 work time increased as sleep time decreased . In addition, the working hours of 8 9 3 healthcare workers are obviously too long. According to the latest survey data in 10 11 12 4 2017, healthcare workers in China's tertiary hospitals work an average of 51.05 hours 13 14 5 per week 40, which is similar to the hours were reported by participants in this study. 15 16 17 6 Hale et al. reported that people working more than 50 hours per week were more 18 For peer review only 19 41 20 7 likely to be short sleepers . Many countries have clear rules for setting limit on the 21 22 8 maximum working hours of healthcare workers. For example, according to The Labor 23 24 25 9 Standards Act of Japan, the total number of working hours should not exceed 40 26 27 10 h/week 42. The European Union specified a maximum of 48 h of work/week and 28 29 30 11 established rest periods for healthcare workers in 2009 43. Although China’s labor law 31 32 33 12 specifies a maximum working time, it is not applicable to the profession of healthcare 34 35 13 workers 44. Currently, there are no special policies on the labor rights of healthcare 36

37 http://bmjopen.bmj.com/ 38 14 workers 44. 39 40 15 Healthy behaviors also play an important role in sleep problems. Our results show that 41 42 43 16 more than three-quarters of medical staff are unable to maintain a regular diet, and 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 irregular diet was significantly associated with sleep problems. Many healthcare 47 48 18 workers do not have enough time to eat during work time, so they have to order a fast 49 50 51 19 food in delivery services or simply delayed their meal. However, short sleep duration 52 53 20 has been associated with higher energy intake, mostly due to increased consumption 54 55 56 21 of saturated fat 45. Additionally, Grant et al. emphasized that for night shift workers, 57 58 46 59 22 eating at night may increase mistakes at work and delay response times . Therefore, 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 the role of regular diet in sleep quality remains controversial. 5 6 7 2 Regularity of bowel movements is also an important factor affecting sleep. Many 8 9 3 medical personnel are unable to form a regular biological schedule for bowel 10 11 12 4 movements because of their heavy workload. This behavior may eventually lead to 13 14 5 abnormal bowel movements or constipation. One et al. reported that constipation may 15 16 17 6 be a risk factor for sleep problems 47. The identification of this risk factor for sleep 18 For peer review only 19 20 7 problems among healthcare workers is a novel contribution to research field. We also 21 22 8 observed that participants who did not engage in regular exercise were more likely to 23 24 25 9 experience sleep problems, which is consistent with previous researches 48. Indeed, 26 27 10 moderate exercise can relieve stress and improve sleep quality. 28 29 30 11 A study funded by the Chinese Medical Doctor Association showed that there are 46 31 32 33 12 cases of Karoshi among Chinese healthcare workers reported between 2013 and 2015, 34 35 13 and most cases of Karoshi occurred in healthcare workers aged 30-39 years 49. 36

37 http://bmjopen.bmj.com/ 38 14 Although healthcare workers in China's top-three hospitals treat many patients, but 39 40 15 few of them recognize their own health. As we found in this study, healthcare workers 41 42 43 16 generally have problems such as lack of sleep, lack of physical exercise, and irregular 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 diet. Studies have shown that chronically exposure to these factors may increase the 47 48 18 risk of chronic diseases among healthcare workers 7 50 51. The phenomenon that 49 50 51 19 healthcare workers do not pay proper attention to their health should be recognized by 52 53 20 the government and society. 54 55 56 21 In summary, we suggest that it is important for healthcare workers to increase their 57 58 59 22 sleep duration and reduce their workload. The effect of chronic pain, especially 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 headache that had on the healthcare workers’ sleep problems are a special area that 5 6 7 2 requires particular attention. Furthermore, healthy behaviors are equally important. 8 9 3 Based on our findings, we make the following recommendations: (1) regular 10 11 12 4 assessment of the sleep quality among healthcare workers should be promoted; (2) 13 14 5 attention should be paid to chronic pain among medical personnel, especially 15 16 17 6 headaches, and regular physical examinations should be provided; (3) reasonable 18 For peer review only 19 20 7 work schedules and night shift work systems should be formulated; and (4) healthy 21 22 8 behaviors should be encouraged, for example sports activities could be organized for 23 24 25 9 healthcare workers (5) expanding the scope of labor law and formulating special 26 27 10 regulations for the labor rights of healthcare workers. 28 29 30 11 Our study has some limitations that must be acknowledged in light of the results. 31 32 33 12 First, information regarding the exposure and outcome was obtained via a self-report 34 35 13 questionnaire, which may introduce report bias in these variables. In particular, we 36

37 http://bmjopen.bmj.com/ 38 14 determined sleep problems via subjective evaluations of the participants. Second, all 39 40 15 participants were from Heilongjiang Province, so the study findings should be 41 42 43 16 generalized to other populations with care. Third, the sample did not include nursing 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 staff, therefore the results cannot be used to characterize all healthcare workers. 47 48 18 Fourth, as it is a cross-sectional study, we cannot establish causal relationships based 49 50 51 19 on the results. 52 53 20 Future research should incorporate validated scales and clinical diagnoses to support 54 55 56 21 the results obtained via self-report. Although we observed several potential factors 57 58 59 22 that were associated with sleep problems, further studies should also explore the 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 34

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 effects of interventions for sleep problems among healthcare workers. In addition, 5 6 7 2 further qualitative and quantitative study on the possible risk to patient’s safety and 8 9 3 the quality of care due to healthcare workers’ poor sleep should be encouraged. 10 11 4 12 13 5 Conclusions 14 15 16 6 Our findings suggest that over 45% of the healthcare workers from Heilongjiang 17 18 7 province had sleepFor problems. peer Headache review was a prominent only risk factor for sleep 19 20 21 8 problems in this population. Long working hours and frequent night shifts were also 22 23 24 9 positively associated with sleep problems among healthcare workers. However, 25 26 10 healthy behaviors can reduce the likelihood of sleep problems. We recommend that 27 28 29 11 there should be a focus on evaluating and improving sleep quality among healthcare 30 31 12 workers. Moreover, future studies should explore the effects of interventions for sleep 32 33 34 13 problems among healthcare workers. 35 36 14

37 http://bmjopen.bmj.com/ 38 39 15 Acknowledgements 40 41 42 16 The authors are very grateful to Miaomiao Zhao, Ding Ding, Linghan Shan for 43 44 17 providing the suggestion to draft the manuscript. We also thank all the participating

45 on September 27, 2021 by guest. Protected copyright. 46 47 18 organizations and participants. 48 49 50 19 51 52 20 Footnotes 53 54 55 21 Funding 56 57 22 This study was supported by the CMB Distinguished Professorship Awards (Grant 58 59 60 23 No. G16916400).

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 34 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 5 6 7 2 Competing interests 8 9 3 The authors declare that they have no competing interests. 10 11 12 4 13 14 5 Contributors 15 16 17 6 HL conducted calculations, analyzed results and drafted the manuscript. QHW and 18 For peer review only 19 20 7 YHH were responsible for the overall design of the research, organized and conducted 21 22 8 the survey. MLJ, NN and LL designed the analyses framework. ZK and LJG revised 23 24 25 9 the paper. JJL, MC, XT and TZ assisted with the literature review and data collection. 26 27 10 All authors approval of the current version of this manuscript for publications. 28 29 30 11 31 32 33 12 Patient consent 34 35 13 Patient consent was obtained for publication 36

37 http://bmjopen.bmj.com/ 38 14 39 40 15 Data sharing statement 41 42 43 16 Please contact author for data requests. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 47 48 18 References 49 50 19 1. Koyanagi A, Stickley A. The Association between Sleep Problems and Psychotic Symptoms in the 51 20 General Population: A Global Perspective. Sleep 2015;38(12):1875-85. doi: 52 53 21 10.5665/sleep.5232 [published Online First: 2015/06/19] 54 22 2. Chinese Medical Association. China Sleep Quality Index announced 2015. 2015. 55 23 3. Hayashino Y, Fukuhara S, Suzukamo Y, et al. Relation between sleep quality and quantity, quality of 56 57 24 life, and risk of developing diabetes in healthy workers in Japan: the High-risk and Population 58 25 Strategy for Occupational Health Promotion (HIPOP-OHP) Study. BMC Public Health 59 26 2007;7:129. doi: 10.1186/1471-2458-7-129 [published Online First: 2007/06/29] 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 34

1 2 3 1 4. Patterson PD, Weaver MD, Frank RC, et al. Association between poor sleep, fatigue, and safety BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 outcomes in emergency medical services providers. Prehosp Emerg Care 2012;16(1):86-97. 6 3 doi: 10.3109/10903127.2011.616261 [published Online First: 2011/10/26] 7 4 5. Sleep deprivation, sleep disorders, fatigue, stress and the risk of occupational injuries and errors 8 9 5 among health-care workers: a multi-approach epidemiological study. 10 6 6. Ghalichi L, Pournik O, M G. Sleep quality among health care workers. Arch Iran Med 11 7 2013;16(2):100-03. 12 8 7. Buxton OM, Marcelli E. Short and long sleep are positively associated with obesity, diabetes, 13 14 9 hypertension, and cardiovascular disease among adults in the United States. Social Science & 15 10 Medicine 2010;71(5):1027-36. 16 11 8. Irwin MR. Why sleep is important for health: a psychoneuroimmunology perspective. Annu Rev 17 18 12 Psychol 2015;66:143-72.For peer doi: 10.1146/annurev-psych-010213-115205 review only [published Online 19 13 First: 2014/07/26] 20 14 9. Gottlieb DJ, Redline S, Nieto FJ, et al. Association of usual sleep duration with hypertension: the 21 22 15 Sleep Heart Health Study. Sleep 2006;29(8):1009-14. 23 16 10. Sun W, Fu J, Chang Y, et al. Epidemiological study on risk factors for anxiety disorder among 24 17 Chinese doctors. Journal of Occupational Health 2012;54(1):1-8. 25 18 11. Patterson PD, Weaver MD, Frank RC, et al. Association Between Poor Sleep, Fatigue, and Safety 26 27 19 Outcomes in Emergency Medical Services Providers. Prehospital Emergency Care 28 20 2011;16(1):86-97. 29 21 12. Eddy R. Sleep deprivation among physicians. British Columbia Medical Journal 2005(4):176-80. 30 31 22 13. Ayas NT, Barger LK, Cade BE, et al. Extended work duration and the risk of self-reported 32 23 percutaneous injuries in interns. JAMA 2006;296(9):1055-62. doi: 10.1001/jama.296.9.1055 33 24 [published Online First: 2006/09/07] 34 35 25 14. Chaudhury PK, Warnock GL, Whalen TV, et al. CAGS and ACS evidence based reviews in surgery. 36 26 Risks of complications by attending physicians after performing nighttime procedures. Can J

37 27 Surg 2012;55(5):337-9. doi: 10.1503/cjs.018212 [published Online First: 2012/09/21] http://bmjopen.bmj.com/ 38 28 15. Mycyk MB, McDaniel MR, Fotis MA, et al. Hospitalwide adverse drug events before and after 39 40 29 limiting weekly work hours of medical residents to 80. Am J Health Syst Pharm 41 30 2005;62(15):1592-5. doi: 10.2146/ajhp040527 [published Online First: 2005/07/21] 42 31 16. Lockley SW, Landrigan CP, Barger LK, et al. When policy meets physiology: the challenge of 43 44 32 reducing resident work hours. Clin Orthop Relat Res 2006;449:116-27. doi: 10.1097/01.blo.0000224057.32367.84 [published Online First: 2006/06/14]

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 17. Department of Economic and Social Affairs Population Division. World population ageing 2015. 47 48 35 United Nations. 2015; 135-154.18. Chen W, Zheng R, Baade PD, et al. Cancer statistics in 49 36 China, 2015. CA Cancer J Clin 2016;66(2):115-32. doi: 10.3322/caac.21338 [published Online 50 37 First: 2016/01/26] 51 38 19. Moran A, Gu D, Zhao D, et al. Future cardiovascular disease in china: markov model and risk factor 52 53 39 scenario projections from the coronary heart disease policy model-china. Circ Cardiovasc 54 40 Qual Outcomes 2010;3(3):243-52. doi: 10.1161/CIRCOUTCOMES.109.910711 [published 55 41 Online First: 2010/05/06] 56 57 42 20. Kobayashi T, Suzuki E, Takao S, et al. Long working hours and metabolic syndrome among 58 43 Japanese men: a cross-sectional study. Bmc Public Health 2012;12(1):395. 59 44 21. ALEMOHAMMAD ZB, TAGHAVI SME, SHARIFIAN A. Sleep Quality and Restless Legs Syndrome 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 34 BMJ Open

1 2 3 1 among Health-care Workers: Shift Workers and Non-shift Workers. Journal of Sleep Sciences BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 2016;1(2):74-79. 6 3 22. Andersen LL, Clausen T, Carneiro IG, et al. Spreading of chronic pain between body regions: 7 4 prospective cohort study among health care workers. Eur J Pain 2012;16(10):1437-43. doi: 8 9 5 10.1002/j.1532-2149.2012.00143.x [published Online First: 2012/03/31] 10 6 23. Jank R, Gallee A, Boeckle M, et al. Chronic Pain and Sleep Disorders in Primary Care. Pain Res Treat 11 7 2017;2017:9081802. doi: 10.1155/2017/9081802 [published Online First: 2018/02/08] 12 8 24. Stonge MP, Mikic A, Pietrolungo CE. Effects of Diet on Sleep Quality. Advances in Nutrition 13 14 9 2016;7(5):938. 15 10 25. National Bureau of Statistics. China health statistical yearbook 2017. Beijing Union Medical 16 11 University Press. 2017; 118:138. 17 18 12 26. Chinese MedicalFor Doctor Association.peer Chinese review doctors practicing only white paper 2015. 19 13 http://www.cmda.net/zlwqgzdt/596.jhtml [Accessed 14 Apr 2018]. 20 14 27. Jafari Roodbandi A, Choobineh A, Daneshvar S. Relationship between circadian rhythm amplitude 21 22 15 and stability with sleep quality and sleepiness among shift nurses and health care workers. 23 16 Int J Occup Saf Ergon 2015;21(3):312-7. doi: 10.1080/10803548.2015.1081770 [published 24 17 Online First: 2015/09/29] 25 18 28. Kolo ES, Ahmed AO, Hamisu A, et al. Sleep health of healthcare workers in Kano, Nigeria. Niger J 26 27 19 Clin Pract 2017;20(4):479-83. doi: 10.4103/1119-3077.204378 [published Online First: 28 20 2017/04/14] 29 21 29. Ertel KA, Berkman LF, Buxton OM. Socioeconomic status, occupational characteristics, and sleep 30 31 22 duration in African/Caribbean immigrants and US White health care workers. Sleep 32 23 2011;34(4):509-18. [published Online First: 2011/04/05] 33 24 30. National Health and Family Planning Commission. China health statistical yearbook 2015. Beijing: 34 35 25 China Union Medical University Press, 2015; 11. 36 26 31. Stranges S, Tigbe W, Gomez-Olive FX, et al. Sleep problems: an emerging global epidemic?

37 27 Findings from the INDEPTH WHO-SAGE study among more than 40,000 older adults from 8 http://bmjopen.bmj.com/ 38 28 countries across Africa and Asia. Sleep 2012;35(8):1173-81. doi: 10.5665/sleep.2012 39 40 29 [published Online First: 2012/08/02] 41 30 32. Chen W, Zhou F, Hall BJ, et al. Is there a relationship between geographic distance and uptake of 42 31 HIV testing services? A representative population-based study of Chinese adults in 43 44 32 Guangzhou, China. PLoS One 2017;12(7):e0180801. doi: 10.1371/journal.pone.0180801 [published Online First: 2017/07/21]

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 33. Ju H, Pan XL, Li A. Multi-level modelling of the factors that influence the participation of disabled 47 48 35 rural individuals in social medical insurance in China. Bmc Health Services Research 49 36 2013;13(1):58-58. 50 37 34. Onwuekwe I, Onyeka T, Aguwa E, et al. Headache prevalence and its characterization amongst 51 38 hospital workers in Enugu, South East Nigeria. Head & face medicine 2014;10(1):48. 52 53 39 35. Holland PR. Headache and sleep: shared pathophysiological mechanisms. Cephalalgia 54 40 2014;34(10):725-44. doi: 10.1177/0333102414541687 [published Online First: 2014/07/24] 55 41 36. Zhang Y, Punnett L, Mcenany GP, et al. Contributing influences of work environment on sleep 56 57 42 quantity and quality of nursing assistants in long-term care facilities: A cross-sectional study. 58 43 Geriatric Nursing 2015;37(1):13-18. 59 44 37. Burgess HJ, Sharkey KM, Eastman CI. Bright light, dark and melatonin can promote circadian 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 34

1 2 3 1 adaptation in night shift workers. Sleep Medicine Reviews 2002;6(5):407-20. BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 38. Neil-Sztramko SE, Pahwa M, Demers PA, et al. Health-related interventions among night shift 6 3 workers: a critical review of the literature. Scandinavian Journal of Work Environment & 7 4 Health 2014;40(6):543-56. 8 9 5 39. Basner M, Fomberstein KM, Razavi FM, et al. American Time Use Survey: Sleep Time and Its 10 6 Relationship to Waking Activities. Sleep 2007;30(9):1085-95. 11 7 40. Chinese Medical Doctor Association. Chinese doctors practicing white paper 2017. 12 8 http://www.cmda.net/u/cms/www/201807/06181247ffex.pdf. [Accessed 20 May 2018] 13 14 9 41. Hale L. Who has time to sleep? J Public Health 2005;27(2):205-11. 15 10 42. Takeuchi M, Rahman M, Ishiguro A, et al. Long working hours and pregnancy complications: 16 11 women physicians survey in Japan. BMC Pregnancy Childbirth 2014;14:245. doi: 17 18 12 10.1186/1471-2393-14-245For peer [published review Online First: 2014/07/26] only 19 13 43. Lancet T. Doctors’ training and the European Working Time Directive. Lancet 20 14 2010;375(9732):2121. 21 22 15 44. Ren Y. Study on Labor Rights Guarantee of Medical Staff in China. Legal System and Society 23 16 2017(4):293-94. 24 17 45. Patel SR, Hu FB. Short sleep duration and weight gain: a systematic review. Obesity 25 18 2008;16(3):643. 26 27 19 46. Grant CL, Dorrian J, Coates AM, et al. The impact of meal timing on performance, sleepiness, 28 20 gastric upset, and hunger during simulated night shift. Industrial Health 2017;55(5):423-36. 29 21 47. Ono S, Komada Y, Kamiya T, et al. A pilot study of the relationship between bowel habits and sleep 30 31 22 health by actigraphy measurement and fecal flora analysis. Journal of Physiological 32 23 Anthropology 2008;27(3):145-51. 33 24 48. Han Y, Yuan Y, Zhang L, et al. Sleep disorder status of nurses in general hospitals and its 34 35 25 influencing factors. Psychiatria Danubina 2016;28(2):176-83. 36 26 49. Shan HP, Yang XH, Zhan XL, et al. Overwork is a silent killer of Chinese doctors: a review of Karoshi

37 27 in China 2013-2015. Public Health 2017;147:98. http://bmjopen.bmj.com/ 38 28 50. Thomas SJ, Calhoun D. Sleep, insomnia, and hypertension: current findings and future directions. J 39 40 29 Am Soc Hypertens 2017;11(2):122-29. doi: 10.1016/j.jash.2016.11.008 [published Online 41 30 First: 2017/01/23] 42 31 51. Yu H, Xie LF, Chen K, et al. Initiating Characteristics of Early-onset Type 2 Diabetes Mellitus in 43 44 32 Chinese Patients. Chin Med J (Engl) 2016;129(7):778-84. doi: 10.4103/0366-6999.178959 [published Online First: 2016/03/22]

45 33 on September 27, 2021 by guest. Protected copyright. 46 47 34 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 34 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 Figure 1. Subjective sleep duration (hours per night) 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 27, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 34 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 Additional file 1. Questionnaire 5 6 ID number：______7 8 [1]Name：______9 10 [2]Age： ______11 ： ① ② 12 [3]Gender _____ Male Female 13 [4]Hospital：______14 15 [5]Department： 16 For peer review only 17 [5.1]Clinical department：______(If you are not a healthcare worker in the clinical 18 19 department, please fill in 0.) 20 21 ①Internal Medicine ②Surgery Department 22 23 ③Obstetricians and Gynecologists ④Pediatrics Department 24 25 ⑤Ophthalmology Department ⑥Stomatology Department 26 27 ⑦Otolaryngology Department ⑧Dermatology Department 28 29 ⑨Anesthesiology Department ⑩Emergency Department 30 31 ⑪Traditional Chinese Medicine Department ⑫Infectious Diseases Department 32 ⑬Plastic Surgery Department ⑭ICU ⑮Others ______33 http://bmjopen.bmj.com/ 34 ： 35 [5.2] Medical technology department ______(If you are not a healthcare worker in 36 the medical technology department, please fill in 0.) 37 38 39 ①Pharmacy Department ②Clinical Lab ③Radiology Department 40

41 ④Pathology Department ⑤Nuclear Medicine Department ⑥Nutritional Department on September 27, 2021 by guest. Protected copyright. 42 43 ⑦Others______44 45 ： 46 [5.3] Administrative department ______(If you are not a healthcare worker in the 47 medical technology department, please fill in 0.) 48 49 ①Personnel Department ②Preventive care department 50 51 ③Medical Service Management Office ④Finance Department 52 53 ⑤Administration Office ⑥Others______54 55 [6]Educational background：______56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 34 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 ①Doctor ②Master ③Bachelor ④Junior college 5 6 ⑤High school or technical secondary school ⑥Others______7 8 [7]Title：______9 10 ①senior title ②intermediate job title ③junior titles ④No title 11 ⑤ 12 Others______13 [8] What is your monthly salary now? ______yuan 14 15 [9] How much time do you work on average per week? ______Hours 16 For peer review only 17 [10] What's your average monthly night shift (over half an hour and above)？ 18 19 ______time(s)/per month 20 21 [11] How much time do you spend on your daily one-way commute? ______22 23 ①<30 minutes ②30-59 minutes ③60-89 minutes ④>=90 minutes 24 25 [12] Was your diet regular at work? ______26 27 ①Yes ②No 28 29 [13] Did you defecate regularly at work? ______30 31 ①Yes ②No 32 [14] Do you have time for physical exercise every day? ______33 http://bmjopen.bmj.com/ 34 ① ② 35 Yes No 36 [15] What is your average time for sleeping every day in the past month？______37 38 ______hour(s) 39 40 [16] Overall in the last 30 days, how much of a problem did you have with sleeping, such

41 on September 27, 2021 by guest. Protected copyright. 42 as falling asleep, waking up frequently during the night, or waking up too early in the 43 44 morning? ______45 46 ①None ②Mild ③Moderate ④Severe ⑤Extreme 47 48 [17]Do you have the following chronic pain problems in the last 30 days？ 49 50 [17.1] Headaches ______①Yes ②No 51 52 [17.2] Neck and shoulder pain ______①Yes ②No 53 54 [17.3] Lumbar and back pain ______①Yes ②No 55 56 [17.4] Knee pain ______①Yes ②No 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open Page 32 of 34

1 2 STROBE 2007 (v4) Statement—Checklist of items that should be included in reports of cross-sectional studies 3 4 5 Item Section/Topic Recommendation Reported on page # 6 # 7 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract Page 3, line 13 8 9 (b) Provide in the abstract an informative and balanced summary of what was done and what was found Page 3, line 14-22; 10 Page 4, line 1-7 11 12 Introduction For peer review only 13 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported Page 5, line 1-22; 14 Page 6, line 1-22; 15 Page 7, line 1-22; 16 17 http://bmjopen.bmj.com/ Page 8, line 1-6 18 Objectives 3 State specific objectives, including any prespecified hypotheses Page 8, line 7-10 19 20 Methods 21 Study design 4 Present key elements of study design early in the paper Page 6, line 16-22; 22 Page 7, line 1-10; 23 Page 9, line 9-10 24 25 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, on September 27, 2021 by guest. Protected copyright. and data Page 8, line 13-23 26 collection Page 9, line 1-6 27 Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of participants Page 8, line 13-23; 28 Page 9, line 8-10 29 30 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if No 31 32 applicable 33 Data sources/ 8* For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe Page 9, line 12-22; 34 measurement comparability of assessment methods if there is more than one group Page 10; 35 Page 11, line 1-20 36 37 Bias 9 Describe any efforts to address potential sources of bias No 38 Study size 10 Explain how the study size was arrived at No 39 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and Page 9, line 12-22; 40 41 why Page 10; 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Page 33 of 34 BMJ Open

1 2 Page 11, line 1-13 3 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding Page 11, line 1-19 4 5 (b) Describe any methods used to examine subgroups and interactions no 6 7 (c) Explain how missing data were addressed Page 9, line 4-6 8 (d) If applicable, describe analytical methods taking account of sampling strategy Page 8, line 20-23; 9 Page 9, line 1-3 10 11 (e) Describe any sensitivity analyses no 12 Results For peer review only 13 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, examined for eligibility, Page 8, line 20-23; 14 confirmed eligible, included in the study, completing follow-up, and analysed Page 9, line 1-6 15 16 (b) Give reasons for non-participation at each stage No 17 (c) Consider use of a flow diagram http://bmjopen.bmj.com/ No 18 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential Page 13, line 14-21 19 20 confounders 21 (b) Indicate number of participants with missing data for each variable of interest No 22 Outcome data 15* Report numbers of outcome events or summary measures Page 13, line 14-21 23 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence Page 15, line 9-14; 24 25 interval). Make clear which confounders were adjusted for and why they were included on September 27, 2021 by guest. Protected copyright. Page 16, line 1-20 26 27 (b) Report category boundaries when continuous variables were categorized Page 10, line 16-22; 28 Page 11, line 1-9; 29 30 Page 11, line 12-20; 31 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period No 32 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses No 33 34 Discussion 35 Key results 18 Summarise key results with reference to study objectives Page 19, line 2-7 36 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and Page 23, line 11-19 37 38 magnitude of any potential bias 39 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from Page 24, line 6-13 40 similar studies, and other relevant evidence 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open Page 34 of 34

1 2 Generalisability 21 Discuss the generalisability (external validity) of the study results Page 23, line 14-16 3 4 Other information 5 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on Page 24, line 21-23 6 which the present article is based 7 8 9 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and unexposed groups in cohort and cross-sectional studies. 10 11 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE 12 checklist is best used in conjunction with this articleFor (freely availablepeer on the Web sitesreview of PLoS Medicine at http://www.plosmedicine.org/, only Annals of Internal Medicine at 13 14 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org. 15 16 17 http://bmjopen.bmj.com/ 18 19 20 21 22 23 24 25 on September 27, 2021 by guest. Protected copyright. 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Sleep problems of healthcare workers in tertiary hospital and influencing factors identified through a multi-level analysis: a cross-sectional study in China. ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-032239.R1

Article Type: Original research

Date Submitted by the 15-Oct-2019 Author:

Complete List of Authors: liu, huan; Harbin Medical University, Liu, jing; Harbin Medical University School of Public Health, Department of Social Medicine chen, mingxi; Harbin Medical University, Social Medicine tan, xiao; Harbin Medical University, Social Medicine zheng, tong; Harbin Medical University, Social Medicine Kang, Zheng; Harbin Medical University, Social Medicine; Gao, Lijun; Harbin Medical University, Social Medicine Jiao, Mingli; Harbin Medical University, Health policy Ning, Ning; Harbin Medical University, Social Medicine Liang, Libo; School of Health Management, Harbin Medical University, Social Medicine; Wu, Qunhong; Harbin Medical University, Social Medicine http://bmjopen.bmj.com/ Hao, Yanhua; Harbin Medical University, Social Medicine

Primary Subject Public health Heading:

Secondary Subject Heading: Health policy, Occupational and environmental medicine, Public health

Healthcare worker, Tertiary hospitals, Sleep problems, Influencing Keywords: factors on September 27, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Title Page 5 6 7 2 8 9 3 Title 10 11 4 Sleep problems of healthcare workers in tertiary hospital and influencing factors 12 5 identified through a multi-level analysis: a cross-sectional study in China. 13 14 6 15 16 7 Author names, affiliations and email addresses 17 1 1 1 1 1 1 18 8 Huan Liu , JingjingFor Liu ,peer Mingxi Chen review, Xiao Tan , Tong only Zheng , Zheng Kang , 19 9 Lijun Gao1, Mingli Jiao2, Ning Ning1, Libo Liang1, Qunhong Wu1*, Yanhua Hao1*. 20 21 10 22 23 1 24 11 Department of Social Medicine, Health Management College, Harbin Medical 25 26 12 University, Harbin, Heilongjiang Province, China. 27 28 2 29 13 Department of Health Policy, Health Management College, Harbin Medical University, 30 31 14 Harbin, Heilongjiang Province, China. 32 33 15 34 35 16 Huan Liu, Department of Social Medicine, Health Management College, Harbin 36 17 Medical University, Harbin, Heilongjiang Province, China. Email:

37 http://bmjopen.bmj.com/ 38 39 18 [email protected] 40 41 19 Jingjing Liu, Department of Social Medicine, Health Management College, Harbin 42 43 20 Medical University, Harbin, Heilongjiang Province, China. Email: 44 21 [email protected]

45 on September 27, 2021 by guest. Protected copyright. 46 47 22 Mingxi Chen, Department of Social Medicine, Health Management College, Harbin 48 49 23 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 50 Xiao Tan, Department of Social Medicine, Health Management College, Harbin 51 24 52 25 Medical University, Harbin, Heilongjiang Province, China. Email: 53 54 26 [email protected] 55 56 27 Tong Zheng, Department of Social Medicine, Health Management College, Harbin 57 58 28 Medical University, Harbin, Heilongjiang Province, China. Email: 59 60 29 [email protected]

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Zheng Kang, Department of Social Medicine, Health Management College, Harbin 5 6 2 Medical University, Harbin, Heilongjiang Province, China. Email: 7 8 3 [email protected] 9 10 4 Lijun Gao, Department of Social Medicine, Health Management College, Harbin 11 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 12 5 13 6 Mingli Jiao, Department of Health Policy, Health Management College, Harbin 14 15 7 Medical University, Harbin, Heilongjiang Province, China. Email: 16 17 8 [email protected] 18 For peer review only 19 9 Ning Ning, Department of Social Medicine, Health Management College, Harbin 20 21 10 Medical University, Harbin, Heilongjiang Province, China. Email: 22 23 11 [email protected] 24 25 12 Libo Liang, Department of Social Medicine, School of Public Health, Harbin Medical 26 27 13 University, Harbin, Heilongjiang Province, China. Email: [email protected] 28 29 14 Qunhong Wu, Department of Social Medicine, Health Management College, Harbin 30 31 15 Medical University, Harbin, Heilongjiang Province, China. Email: 32 33 16 [email protected] 34 35 17 Yanhua Hao, Department of Social Medicine, Health Management College, Harbin 36 18 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected]

37 http://bmjopen.bmj.com/ 38 39 19 40 41 20 *Corresponding authors: 42 43 21 Qunhong Wu 44 22 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District,

45 on September 27, 2021 by guest. Protected copyright. 46 47 23 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 48 49 24 87502851; fax: + 86-0451-87502853 50 25 51 52 26 Yanhua Hao 53 54 27 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District, 55 56 28 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 57 58 29 87502860; fax: + 86-0451-87502853 59 60 30

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Keywords: Healthcare worker; Tertiary hospitals; Sleep problems; Influencing factors 5 6 2 Word counts 7 8 3 word count of the abstract: 300 9 4 word count of manuscript: 4181 10 11 5 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Sleep problems of healthcare workers in tertiary hospital and 5 6 7 2 influencing factors identified through a multi-level analysis: a 8 9 3 cross-sectional study in China. 10 11 12 4 13 14 15 5 Abstract 16 17 6 Objectives: Sleep problems are common among healthcare workers, especially among 18 For peer review only 19 20 7 those who working in tertiary hospitals; however, studies focusing on the healthcare 21 22 8 workers in tertiary hospitals in China are limited. Therefore, we aimed to examine the 23 24 25 9 prevalence of the sleep problem and identify factors contribute to it in tertiary hospitals 26 27 10 of China. 28 29 30 11 Design: A cross-sectional questionnaire survey study. 31 32 33 12 Setting: The study was conducted in 377 departments at 33 tertiary hospitals in 34 35 13 Heilongjiang Province, China. 36

37 http://bmjopen.bmj.com/ 38 14 Participants: The study included a representative sample of 4,009 healthcare workers, 39 40 15 including clinicians, medical technicians, and administrative staff, who were selected 41 42 43 16 using a multi-stage cluster sampling method. Ultimately, 3,810 participants completed 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 valid questionnaires, yielding a response rate of 95%. 47 48 18 Primary and secondary outcome measures: Participants completed a structured 49 50 51 19 questionnaire, which collected data regarding demographics, sleep problems, 52 53 20 employment, chronic pain, and healthy behaviors. A two-level logistic regression 54 55 56 21 model was constructed to examine determinants of sleep problems. 57 58 59 22 Results: We found that nearly half of the healthcare workers reported experiencing 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 sleep problems. The two-level logistic model suggested a positive association of sleep 5 6 7 2 problems with headaches (OR = 2.64, 95% CI = 2.27–3.07), working hours (OR = 1.41, 8 9 3 95% CI = 1.20–1.66), and night shifts (OR = 1.43, 95% CI = 1.21–1.69). In contrast, a 10 11 12 4 regular diet (OR = 0.44, 95% CI = 0.36–0.53) and exercise (OR = 0.65, 95% CI = 0.53– 13 14 5 0.80) protected against sleep problems in healthcare workers. 15 16 17 6 Conclusions: Sleep problems are prominent among healthcare workers. The risk factor 18 For peer review only 19 20 7 most strongly associated with sleep problems among healthcare workers was headaches. 21 22 8 To improve sleep problems in healthcare workers, medical institutions and health 23 24 25 9 policy makers should provide workers with reasonable work schedules and night-shift 26 27 10 work systems, while the healthcare workers themselves should focus on their chronic 28 29 30 11 pain problems and engaging in regular diet and exercise. 31 32 33 12 34 35 13 Strengths and limitations of this study 36

37 http://bmjopen.bmj.com/ 38 14  This was a multi-center study with a large sample size; specifically, we recruited 39 40 15 3,810 healthcare workers from 377 departments at 33 tertiary hospitals in 41 42 43 16 Heilongjiang Province, China, using multi-stage stratified random sampling. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17  Considering the hierarchical structure of the study sample, a two-level logistic 47 48 18 model was applied to examine determinants of sleep problems in order to obtain 49 50 51 19 more accurate and reliable results. 52 53 20  Although the results of the study may not be generalizable to healthcare workers in 54 55 56 21 other countries, the findings likely accurately reflect the sleep problems of many 57 58 59 22 Chinese healthcare workers, particularly those working in tertiary public hospitals 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 in China. 5 6 7 2  Information regarding the exposure and outcome was obtained via a self-report 8 9 3 questionnaire, which may introduce report bias in these variables. 10 11 12 4  The cross-sectional design limited our ability to establish causal relationships based 13 14 5 on the results. 15 16 17 6 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Introduction 5 6 7 2 Sleep problems are increasingly becoming a public health concern. Globally, it is 8 9 3 estimated that 7.6% of the population suffers from sleep problems 1. Common sleep 10 11 12 4 problems include difficulty falling asleep, frequent nighttime awakenings, parasomnias, 13 14 5 noisy breathing, snoring, and obstructive sleep apnea, among others 2. In China, 15 16 17 6 workers in a variety of occupations are at risk of sleep problems, although sleep 18 For peer review only 19 3 20 7 problems among healthcare workers are particularly evident . Many scholars have 21 22 8 reported a high prevalence of sleep problems among Chinese healthcare workers, 23 24 25 9 ranging from 21% to 65.5% 4-6. In fact, it has been shown that sleep problems are 26 27 10 common among healthcare workers worldwide, including clinicians, nurses, medical 28 29 30 11 technicians, and administrative staff. 31 32 33 12 For example, Hayashino et al. observed that 30.7% of Japanese healthcare workers 34 35 13 reported having difficulty initiating sleep 7. Moreover, a study performed in Italy found 36

37 http://bmjopen.bmj.com/ 38 14 that the prevalence of sleep disturbances among healthcare workers was extremely high, 39 40 15 even among healthy controls in the study, with 24.6% of workers reporting difficulties 41 42 43 16 falling asleep and 31.7% experiencing frequent awakenings during the night 8. A study 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 in Iran observed that almost 43% of healthcare workers reported different degrees of 47 48 18 sleep problems 9. 49 50 51 19 It is crucial to pay special attention to sleep problems among healthcare workers for 52 53 20 several reasons. First, sleep problems can affect the health and safety of these workers. 54 55 56 21 For instance, various studies have confirmed that sleep problems have a powerful 57 58 10-12 59 22 influence on the risk of cardiovascular disease, immune disorders, and cancer . Sun 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 and colleagues reported a significant negative correlation between anxiety and sleep 5 6 4 7 2 problems in healthcare workers . Moreover, Kling et al. reported that sleep problems 8 9 3 are associated with an increased risk of work injury 13. Second, sleep problems not only 10 11 12 4 affect the health of the healthcare workers themselves, but may also affect patient safety 13 14 5 and the quality of medical services the healthcare workers provide. Indeed, there is clear 15 16 17 6 evidence indicating that sleep deprivation caused by long work hours can affect both 18 For peer review only 19 14 20 7 the health of healthcare workers and the safety of patients . Further, sleep deprivation, 21 22 8 long work hours, and working the night shift have been found to lead to medical errors, 23 24 25 9 such as increases in the rates of surgical complications, needlestick injuries, adverse 26 27 10 drug events, and diagnostic errors 15-18. 28 29 30 11 Currently, given the increasing size of the aging population 19 and the elevated 31 32 20 21 33 12 incidences of chronic diseases such as cancer and cardiovascular diseases , there is 34 35 13 tremendous pressure on the healthcare system in China. In response to the increasing 36

37 http://bmjopen.bmj.com/ 38 14 demands on health services, the workloads and hours worked by healthcare workers 39 40 15 have likewise increased, which may also lead to more frequent sleep problems in these 41 42 43 16 individuals. One prior study of healthcare workers indicated that long working hours 44

45 22 on September 27, 2021 by guest. Protected copyright. 46 17 were the critical determinant of sleep health . However, few studies have focused on 47 48 18 the relationship between work hours and sleep problems in healthcare workers in China. 49 50 51 19 Moreover, Kim et al. suggested that frequent night shifts are likely to be an important 52 53 20 factor affecting sleep problems among healthcare workers 23. 54 55 56 21 Furthermore, one common issue among healthcare workers, chronic pain 24, has been 57 58 25 59 22 shown to be related to sleep problems , although it remains unclear which specific site 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 of chronic pain has the greatest effect on sleep problems. Healthy behaviors are also an 5 6 7 2 import potential influencing factor. For example, research has shown that unhealthy 8 9 3 diets are associated with sleep problems, while a healthy diet can effectively improve 10 11 12 4 sleep quality 26. A recent study conducted in Japan observed that healthcare workers 13 14 5 who reported having trouble initiating sleep consumed more alcohol and exercised less 15 16 17 6 7. In addition, scholars have noted that sleep health should be measured from many 18 For peer review only 19 27 28 20 7 aspects, with sleep duration being one of the most important dimensions . Although 21 22 8 sleep needs are different between individuals due to genetic and environmental factors 23 24 25 9 29, shorter sleep duration may increase the likelihood of sleeping problems 30. 26 27 10 In 2009, China launched a nationwide systematic reform of healthcare services. 28 29 30 11 Although the reform was successful in some areas, additional steps are required to 31 32 33 12 reform the delivery of China’s healthcare services. Particularly, primary healthcare 34 35 13 facilities have not been able to perform a gate-keeping function and most patients in 36

37 http://bmjopen.bmj.com/ 38 14 China prefer to visit tertiary hospitals, especially tertiary public hospitals, even for 39 40 15 common and minor illnesses. In 2016, there were about 3.27 billion patient visits to 41 42 43 16 Chinese hospitals, 49.8% of which were visits to tertiary hospitals 31. A survey of the 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 Chinese Medical Doctor Association in 2014 showed that 92% of healthcare workers 47 48 18 in tertiary hospitals had to work overtime, and 72% of the healthcare workers who work 49 50 51 19 >60 h a week on average are in tertiary hospitals in China 32. Therefore, sleep problems 52 53 20 of healthcare workers in tertiary hospitals is of paramount importance. Nevertheless, 54 55 56 21 population-based prevalence data on sleep problems among healthcare workers in 57 58 59 22 tertiary hospitals in China are limited. Further, the relationships among sleep problems, 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 work factors, healthy behaviors, and chronic pain remain to be discovered in this 5 6 7 2 population. 8 9 3 Therefore, the purposes of the present study were to investigate the prevalence of self- 10 11 12 4 reported sleep problems among healthcare workers in tertiary public hospitals in China 13 14 5 and to identify potential factors associated with sleep problems among these individuals. 15 16 17 6 This study has several advantages. First, although people have realized that sleep 18 For peer review only 19 20 7 problems in healthcare workers are common, studies on healthcare workers in tertiary 21 22 8 hospitals in China are limited. Second, considering the hierarchical structure of the 23 24 25 9 study sample, a two-level logistic model was applied to examine determinants of sleep 26 27 10 problems in order to obtain more accurate and reliable results. Third, compared to 28 29 30 11 previous studies 33-35, our study is a multi-center study that covers a large sample of 31 32 33 12 healthcare workers in the Heilongjiang Province of China. 34 35 13 36 14 Methods

37 http://bmjopen.bmj.com/ 38 39 15 Data collection and study population 40 41 42 16 This study was undertaken in Heilongjiang Province, in northeast China, where there 43 44 17 were 689 public hospitals in 2014, of which 82 were tertiary public hospitals 36. This

45 on September 27, 2021 by guest. Protected copyright. 46 47 18 study focused on healthcare workers in representative tertiary hospitals. The study 48 49 50 19 participants included clinicians (physicians, surgeons, pediatricians, gynecologists, 51 52 20 emergency doctors, etc.), medical technicians (pharmacists, radiological technicians, 53 54 55 21 laboratory physicians, etc.), and office workers (administrative department personnel, 56 57 22 finance personnel, etc.). In this study, we employed a multi-stage cluster sampling 58 59 60 23 method to obtain a representative sample of healthcare workers from tertiary hospitals

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 in China. First, 33 tertiary public hospitals were selected at random to serve as survey 5 6 7 2 locations for this study and all selected hospitals participated (100% response rate). 8 9 3 Second, the departments of each hospital were coded and several departments were 10 11 12 4 randomly selected from each hospital. The sample size at each level was based on the 13 14 5 proportion of the number of workers in the hospital at that level. In the final sample, 15 16 17 6 4,007 healthcare workers from 317 departments were invited to participate in the study. 18 For peer review only 19 20 7 A self-administered questionnaire was completed by the healthcare workers in the 21 22 8 selected departments (see Additional file 1 for details). After the data cleaning process, 23 24 25 9 incomplete or anomalous data (n = 197) were excluded, leaving a final sample of 3,810 26 27 10 participants (95% response rate). 28 29 30 11 31 32 33 12 Variables 34 35 13 A structured questionnaire was designed to obtain data regarding demographics, sleep 36

37 http://bmjopen.bmj.com/ 38 14 problems, working conditions, chronic pain, and healthy behaviors. 39 40 15 41 42 43 16 Outcome measure (dependent variable) 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 The main outcome was the presence of sleep problems, which was a binary variable (1 47 48 18 = yes; 0 = no). The presence of sleep problems was assessed from each participant’s 49 50 51 19 self-report. Participants were asked the question “Overall, in the last 30 days, how much 52 53 20 of a problem did you have with sleeping, such as falling asleep, waking up frequently 54 55 56 21 during the night, or waking up too early in the morning?” with the following five answer 57 58 59 22 options: none, mild, moderate, severe, and extreme. Participants who answered extreme 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 and severe were considered to have sleep problems. This definition has been used in 5 6 1 37 7 2 previous publications that used the same survey question to assess sleep problems . 8 9 3 10 11 12 4 Individual-level variables (independent variables) 13 14 5 Individual-level variables included sociodemographic indicators (age, sex, educational 15 16 17 6 background, job title, and monthly salary income), working conditions (average weekly 18 For peer review only 19 20 7 working hours, number of night shifts per month, and one-way commute time), healthy 21 22 8 behaviors (physical exercise and regularity of diet), and experience of chronic pain 23 24 25 9 (headache, neck and shoulder pain, lumbar and back pain, and knee pain). 26 27 10 Age was categorized into three groups (<30, 30–40, and >40 years) and job title was 28 29 30 11 considered an ordinal categorical variable (junior titles and others, intermediate job title, 31 32 33 12 and senior title). The median monthly salary income of participants in this study was 34 35 13 3,000 yuan. Therefore, we coded monthly salary income as ≤3,000 yuan (=0) 36

37 http://bmjopen.bmj.com/ 38 14 and >3,000 yuan (=1). Average weekly working hours were evaluated with the question: 39 40 15 “How much time do you work on average per week?” The median number of weekly 41 42 43 16 working hours reported by participants in this study was 55 h. Hence, the response 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 categories were 1 = ≥55 h and 0 = <55 h. 47 48 18 Herein, chronic pain was defined as any continuous or persistent intermittent pain 49 50 51 19 experienced for a period longer than 3 months 38. We asked participants whether they 52 53 20 had experienced chronic pain that had not improved in the past 30 days in the following 54 55 56 21 regions: headache, neck and shoulder, lumbar and back, and knee. The response 57 58 59 22 categories were 1 = yes and 0 = no. The coding of other individual-level variables is 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 shown in Additional file 2. 5 6 7 2 8 9 3 Department-level variables (independent variable) 10 11 12 4 The participants’ department type was a level-2 explanatory variable in our analysis 13 14 5 because individuals in the same department are not independent of each other. Among 15 16 17 6 the healthcare workers in the study, some practitioners were from clinical departments 18 For peer review only 19 20 7 (including internal medicine, surgery, gynecology, pediatrics, etc.) and others were 21 22 8 health technicians (including laboratory, pathology, radiology, etc.). Moreover, some 23 24 25 9 health personnel from administrative departments participated in the survey. Therefore, 26 27 10 this study is based on the functional attributes of the department, and whether the 28 29 30 11 participants were from a clinical department (1 = yes, 0 = no) was an explanatory 31 32 33 12 variable. 34 35 13 36

37 http://bmjopen.bmj.com/ 38 14 Data analysis 39 40 15 In this study, we used two statistical methods. The 2-test was used to examine the 41 42 43 16 independent associations between participants’ sleep problems and each independent 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 variable. Since the data were hierarchical, participants (level-1) were nested within 47 48 18 departments (level-2), to create a two-level logistic model that was applied to assess 49 50 51 19 associations between sleep problems and study variables. 52 53 20 Intra-class correlation coefficients (ICC) for healthcare workers were calculated to 54 55 56 21 assess clustering by department 39. First, an empty model was developed to verify 57 58 59 22 whether our research model was applicable to the two-level logistic model. If the ICC 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 was significant, then it suggested that a multi-level logistic regression model was 5 6 7 2 appropriate. Subsequently, model 1 added the department-level explanatory variable to 8 9 3 the empty model. Next, model 2 added individual-level explanatory variables to model 10 11 12 4 1. Meanwhile, we needed to judge whether the two multi-category variables (age and 13 14 5 job title) should be introduced into the model as dummy variables or grouped linear 15 16 17 6 variables according to the fit of model. The likelihood ratio (LR) chi-square test can be 18 For peer review only 19 20 7 used for model comparison and the difference in -2 log-likelihood between two models 21 22 8 are used for the LR chi-square test 40. A statistically significant difference in −2 log- 23 24 25 9 likelihood between the two models indicated that the fit of the latter model was better 26 27 10 41. We also conducted subpopulation analyses to determine if participants with short 28 29 30 11 sleep durations (Q15: 1 = sleep duration ≤6 h, 0 = sleep duration more than 6h) 42, 31 32 43 33 12 significantly modified the observed associations . 34 35 13 The two-level binary logistic regression models were estimated with the GLIMMIXED 36

37 http://bmjopen.bmj.com/ 38 14 and NLMIXED procedures of SAS 9.4 (SAS Institute Inc., Cary, NC, USA). All other 39 40 15 analyses were conducted in SPSS 21.0 (IBM Corp., Armonk, NY, USA), and p values 41 42 43 16 <0.05 were considered to reflect statistical significance. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 47 48 18 Ethics approval 49 50 51 19 The study protocol was reviewed and approved by the Research Ethics Committee of 52 53 20 Harbin Medical University (project identification code: HMUIRB20170016) and the 54 55 56 21 ethics committee from each hospital (see the Additional file 3). The study also 57 58 59 22 received permission and assistance from each hospital to contact their healthcare 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 workers. All participants were informed that the study was voluntary and confidential. 5 6 7 2 Participants were informed that they have the right to discontinue their participation 8 9 3 or withdraw their consent at any time and were not obliged to state their reasons. 10 11 12 4 After participants provided written informed consent, interviews were conducted. 13 14 5 Survey data were collected using an anonymous questionnaire to protect the privacy 15 16 17 6 of participants. 18 For peer review only 19 20 7 21 22 8 Patient and public involvement 23 24 25 9 No patients were involved. 26 27 28 10 29 30 11 Results 31 32 33 12 Participant characteristics 34 35 13 Participants’ general characteristics are shown in Table 1. Overall, participants were 36

37 http://bmjopen.bmj.com/ 38 14 generally unable to maintain a regular diet (75.5%). The most common types of chronic 39 40 15 pain were neck and shoulder pain (69.9%), and lumbar and back pain (64.1%). Sleep 41 42 43 16 problems were reported by 48.5% of participants. The univariate analysis indicated that 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 demographics (excluding sex and monthly salary income), chronic pain, and working 47 48 18 conditions were all related to participants’ sleep problems (p < 0.05). 49 50 51 19 52 53 20 Table 1. Characteristics of questionnaire respondents and results of the univariate 54 21 analysis 55 Variable Total Sleep problems p-value 56 57 N = 3810 (%) No Yes 58 59 n =1964 (%) n = 1846 (%) 60 Department-level

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 43

1 2 3 Clinical department .001 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Yes 3,297 (86.5) 1,658 (50.2) 1,639 (49.7) 6 No 513 (13.5) 306 (59.6) 207 (40.4) 7 Individual-level 8 9 Age (years) < .001 10 < 30 504 (13.2) 308 (61.1) 196 (38.9) 11 30–40 2,054 (53.9) 982 (47.8) 1,072 (52.2) 12 13 > 40 1,252 (32.9) 674 (53.8) 578 (46.2) 14 Sex .279 15 Men 2,455 (64.4) 1,282 (52.2) 1,173 (47.8) 16 17 Women 1,355 (35.6) 682 (50.3) 673 (49.7) 18 Educational backgroundFor peer review only .004 19 20 Bachelor degree and below 1,883 (49.4) 926 (49.2) 957 (50.8) 21 Master's degree and above 1,927 (50.6) 1,038 (53.9) 889 (46.1) 22 23 Job title .001 24 Junior titles and others 924 (24.3) 521 (48.8) 403 (51.2) 25 Intermediate title 1,287 (33.8) 622 (30.8) 665 (69.2) 26 27 Senior title 1,599 (42.0) 821 (51.3) 778 (48.7) 28 Monthly salary income .791 29 ≤ 3,000 yuan 2,286 (60.0) 1,112 (48.6) 1,174 (51.4) 30 31 > 3,000 yuan 1,524 (40.0) 790 (51.8) 734 (48.2) 32 One-way commute time .001 33 < 30 minutes 2,038(53.5) 1097 (53.8) 941 (46.2) 34 35 ≥ 30 minutes 1,772(46.5) 867 (48.9) 905 (51.1) 36 Regular diet at work < .001

37 Yes 934 (24.5) 682 (73.0) 252 (27.0) http://bmjopen.bmj.com/ 38 39 No 2,876 (75.5) 1,282 (44.6) 1,594 (55.4) 40 No 3,063 (80.4) 1,427 (34.2) 1,636 (65.8) 41 42 Work hours per week (hours) < .001 43 ≥ 55 1,964 (51.5) 850 (42.3) 1,114 (56.7) 44 < 55 1,846 (48.5) 1,114 (60.3) 732 (39.7)

45 on September 27, 2021 by guest. Protected copyright. 46 Night shifts per month (times) < .001 47 ≤ 5 2,171 (57.0) 1,270 (58.5) 901 (41.5) 48 49 > 5 1,639 (43.0) 694 (42.3) 945 (57.7) 50 Daily exercise < .001 51 52 Yes 685 (18.0) 452 (66.0) 233 (34.0) 53 No 3,125 (82.0) 1,512 (48.4) 1,613 (51.6) 54 Headaches < .001 55 56 Yes 1,698 (44.6) 601 (35.4) 1,097 (64.6) 57 No 2,112 (55.4) 1,363 (64.5) 749 (35.5) 58 59 Neck and shoulder pain < .001 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 43 BMJ Open

1 2 3 Yes 2,664 (69.9) 1,243 (46.7) 1,421 (53.3) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 No 1,146 (30.1) 721 (62.9) 425 (37.1) 6 Lumbar and back pain < .001 7 8 Yes 2,441 (64.1) 1,111 (45.5) 1,330 (54.5) 9 No 1,369 (35.9) 853 (62.3) 516 (37.7) 10 11 Knee pain < .001 12 Yes 975 (25.6) 340 (34.9) 635 (65.1) 13 No 2,835 (74.4) 1,624 (57.3) 1,211 (42.7) 14 15 1 16 17 2 Self-reported sleep duration 18 For peer review only 19 20 3 Figure 1 shows that the largest proportion of respondents reported sleeping for about 21 22 4 6 h, regardless of department. The largest proportion of healthcare workers who 23 24 25 5 reported sleeping ≤5 h were those working in clinical departments. Moreover, 46.7% 26 27 6 of participants reported a sleep duration about 6 hours per day and 11.3% less than 5 28 29 30 7 hours. 31 32 33 8 34 35 9 Multi-level analysis 36

37 http://bmjopen.bmj.com/ 38 10 Table 2 presents the results from the multi-level analyses. The results of running an 39 40 11 empty model showed significant between-group variation (ICC = 0.1181, p < 0.001). 41 42 43 12 An ICC of 0.1181 indicated that about 12% of the total variation in the outcome 44

45 on September 27, 2021 by guest. Protected copyright. 46 13 variables was caused by variation between the departments. 47 48 14 Subsequently, the department-level explanatory variable was introduced. The results 49 50 51 15 showed that the odds of having sleep problems was about 1.5 times greater among 52 53 16 healthcare workers in clinical departments than among those from non-clinical 54 55 56 17 departments (OR = 1.53, 95% CI = 1.15–2.04). After incorporating the department- 57 58 59 18 level explanatory variable, the −2 log-likelihood declined from 5,159.4 to 5,150.9. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 The difference in −2 log-likelihood between model 1 and the empty model indicated 5 6 7 2 that the fit of model 1 was better (LR = 8.5, df = 1, p < 0.01). 8 9 3 In model 2, individual-level variables were added. The results showed that sleep 10 11 12 4 problems for healthcare workers had a protective association with regular diet (OR = 13 14 5 0.44, 95% CI = 0.36–0.53) and exercise (OR = 0.65, 95% CI = 0.53–0.80). 15 16 17 6 Participants with junior titles had lower odds of suffering from sleep problems than 18 For peer review only 19 20 7 did participants with a senior job title (OR = 0.71, 95% CI = 0.55–0.92). Model 2 21 22 8 further suggested that chronic pain was positively associated with sleep problems. 23 24 25 9 Specifically, headaches were more likely to be reported by participants with sleep 26 27 10 problems (OR = 2.64, 95% CI = 2.27–3.03). Furthermore, working hours (OR = 1.41, 28 29 30 11 95% CI = 1.20–1.66) and night shifts (OR = 1.43, 95% CI = 1.21–1.69) were 31 32 33 12 associated with sleep problems. The difference in −2 log-likelihood between model 2 34 35 13 and model 1 indicated that the fit of model 2 was better (LR = 612, df = 14, p < 36

37 http://bmjopen.bmj.com/ 38 14 0.001). Meanwhile, department-level variables were also tested but showed no 39 40 15 statistical significance. 41 42 16 43 44 17 Table 2. Factors related to sleep problems: results from the two-level logistic regression

45 18 analysis (N = 3,810) on September 27, 2021 by guest. Protected copyright. 46 Variable Empty model Model 1 Model 2 47 48 Department-level 49 Clinical department 50 Yes 1.53 (1.15, 2.04) ** 1.07 (0.81, 1.40) 51 52 No 1 (reference) 1 (reference) 53 Individual-level 54 Age (years) 55 < 30 0.87 (0.63, 1.20) 56 57 30–40 1.06 (0.86, 1.31) 58 > 40 1 (reference) 59 60 Educational background

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 43 BMJ Open

1 2 3 Bachelor degree and below 0.87 (0.74, 1.03) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Master's degree and above 1 (reference) 6 Job title 7 Junior title or no title 0.71 (0.55, 0.92) ** 8 9 Intermediate title 0.87 (0.71, 1.07) 10 Senior title 1 (reference) 11 One-way commute time 12 13 <30 minutes 1.07 (0.92, 1.24) 14 ≥30 minutes 1 (reference) 15 Regular diet at work 16 Yes 0.44 (0.36, 0.53) *** 17 18 No For peer review only 1 (reference) 19 Daily exercise 20 Yes 0.65 (0.53, 0.80) *** 21 22 No 1 (reference) 23 Work hours per week 24 ≥ 55 1.41 (1.20, 1.66) *** 25 26 < 55 1 (reference) 27 Night shifts per month 28 ≤ 5 1.43 (1.21, 1.69) *** 29 > 5 1 (reference) 30 31 Headache 32 Yes 2.64 (2.27, 3.07) *** 33 No 1 (reference) 34 35 Neck and shoulder pain 36 Yes 1.29 (1.09, 1.52) **

37 No 1 (reference) http://bmjopen.bmj.com/ 38 39 Lumbar and back pain 40 Yes 1.36 (1.16, 1.60) *** 41 No 1 (reference) 42 43 Knee pain 44 Yes 1.61 (1.34, 1.93) ***

45 No 1 (reference) on September 27, 2021 by guest. Protected copyright. 46 47 Sigma μ (SE) 0.4407 (0.092) 0.4061 (0.088) 0.2278 (0.070) 48 Intra-class correlation 0.1181*** 0.1099*** 0.0648*** 49 −2 log-likelihood 5159.4 5150.9 4538.0 50 * p < 0.05; ** p < 0.01; *** p < 0.001 51 1 52 53 2 54 55 3 Table 3 shows the results from the two-level logistic regression model for the 56 57 58 4 subpopulation of healthcare workers with short sleep durations. In this subpopulation, 59 60 5 the observed associations were mostly consistent with those identified with the two-

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 level model comprising the total sample. However, in contrast to the analysis of the 5 6 7 2 total sample, neck and shoulder pain (p = 0.128) was not significantly associated with 8 9 3 sleep problems in healthcare workers with short sleep durations. 10 11 12 4 13 5 Table 3. Factors related to sleep problems: results from the two-level logistic regression 14 6 analysis of healthcare workers with short sleep durations (≤6 h; n = 2,210) 15 16 Variable Empty model Model 1 Model 2 17 Department-level 18 Clinical departmentFor peer review only 19 * 20 Yes 1.47(1.21, 1.79) 0.86 (0.59, 1.25) 21 No 1 (reference) 1 (reference) 22 Individual-level 23 24 Age (years) 25 < 30 0.92 (0.60, 1.42) 26 30–40 1.11 (0.84, 1.46) 27 28 > 40 1 (reference) 29 Educational background 30 Bachelor degree and below 0.86 (0.69, 1.06) 31 32 Master's degree and above 1 (reference) 33 Job title 34 Junior title or no title 0.63 (0.45, 0.88) ** 35 36 Intermediate title 0.84 (0.63, 1.11) Senior title 1 (reference)

37 http://bmjopen.bmj.com/ 38 One-way commute time 39 <30 minutes 1.13 (0.93, 1.38) 40 41 ≥30 minutes 1 (reference) 42 Regular diet at work 43 Yes 0.53 (0.40, 0.71) *** 44 No 1 (reference)

45 on September 27, 2021 by guest. Protected copyright. 46 Daily exercise 47 Yes 0.74 (0.56, 0.98) * 48 49 No 1 (reference) 50 Work hours per week 51 ≥ 55 1.32 (1.06, 1.63) *** 52 < 55 1 (reference) 53 54 Night shifts per month 55 ≤ 5 1.45 (1.17, 1.81) ** 56 > 5 1 (reference) 57 58 Headache 59 Yes 2.58 (2.11, 3.16) *** 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 43 BMJ Open

1 2 3 No 1 (reference) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Neck and shoulder pain 6 Yes 1.19 (0.95, 1.49) 7 No 1 (reference) 8 9 Lumbar and back pain 10 Yes 1.29 (1.04, 1.60) * 11 12 No 1 (reference) 13 Knee pain 14 Yes 1.63 (1.29, 2.05) *** 15 No 1 (reference) 16 17 Sigma μ (SE) 0.3885 (0.100) 0.3842 (0.099) 0.2789 (0.098) 18 Intra-class correlationFor peer0.1056 review*** 0.1046 only*** 0.07814*** 19 −2 log-likelihood 2922.1 2920.5 2645.1 20 * ** *** 21 1 p < 0.05; p < 0.01; p < 0.001 22 23 2 24 25 3 Factors related to headaches 26 27 28 4 Table 4 shows factors related to headaches among respondents with sleep problems. 29 30 5 We found that 62.4% of participants with headache worked ≥55 h per week and that 31 32 33 6 73.5% of respondents with headaches were not exercising every day. 34 35 36 7

37 http://bmjopen.bmj.com/ 38 8 Table 4. Factors related to headaches among respondents with sleep problems (n = 39 40 41 9 1,846) 42 43 Variable Headache p-value 44 Yes No

45 on September 27, 2021 by guest. Protected copyright. 46 n= 1097 (%) n= 749 (%) 47 Work hours per week .015 48 ≥55 hours 685(62.4) 429(57.3) 49 50 <55 hours 412(37.6) 320(42.7) 51 Number of night shifts per .173 52 53 month 54 5 times or less 525(47.9) 376(50.2) 55 More than 5 times 572(52.1) 373(49.8) 56 57 Do exercise every day < .001 58 Yes 120 (10.9) 113(15.1) 59 60 No 977 (89.1) 636(84.9)

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 43

1 2 3 Regular diet at work < .001 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Yes 122 (11.12) 130 (17.36) 6 No 975 (88.88) 619 (82.64) 7 8 Neck and shoulder pain < .001 9 Yes 880 (80.2) 540 (72.1) 10 11 No 217 (19.8) 209 (27.9) 12 Lumbar and back pain < .001 13 Yes 833 (75.9) 496 (66.2) 14 15 No 264 (24.1) 253 (33.8) 16 Knee pain < .001 17 18 YesFor peer review498 (45.4) only136 (18.2) 19 No 559 (54.6) 613 (81.8) 20 21 1 22 23 24 2 Discussion 25 26 27 3 This study contributes to current knowledge of the seriousness of sleep problems for 28 29 4 healthcare workers by reporting the prevalence of sleep problems among healthcare 30 31 32 5 workers in tertiary public hospitals in Heilongjiang Province, China, and identifying 33 34 6 factors related to the sleep problems of healthcare workers. Our results show that 35 36

37 7 more than 45% of healthcare workers reported experiencing sleep problems, which is http://bmjopen.bmj.com/ 38 39 40 8 much higher than the prevalence reported for the general population of China (36.2%) 41 42 9 3. 43 44 10 Herein, the factor most strongly associated with sleep problems among healthcare

45 on September 27, 2021 by guest. Protected copyright. 46 47 11 workers was headaches. We found that 44.6% of participants reported experiencing 48 49 50 12 headaches longer than three months and had not improved in the past 30 days. Such 51 52 53 13 findings are consistent with previous studies showing that headache is a very common 54 55 14 symptom among healthcare workers. Sokolovic et al. found that 61% of the healthcare 56 57 58 15 workers they evaluated had experienced at least one type of headache in the previous 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 3 months. A cross-sectional study of healthcare workers conducted in Enugu, South 5 6 7 2 East Nigeria, showed that 12% of the participants had severe headaches that affected 8 9 3 their sleep 44. From a clinical perspective, it has been suggested that the relationship 10 11 12 4 between headaches and sleep problems may be bidirectional 45, whereby the treatment 13 14 5 of headaches among medical personnel is crucial for relieving sleep problems, while 15 16 17 6 the improvement of sleep is also likely to reduce headache symptoms in healthcare 18 For peer review only 19 20 7 workers. In addition, we analyzed factors related to the participants’ headaches and 21 22 8 found that among respondents with sleep problems, an irregular diet, working 23 24 25 9 overtime, a lack of exercise, neck and shoulder pain, lumbar and back pain, and knee 26 27 10 pain were associated with headaches. Such findings imply that additional attention is 28 29 30 11 required to address these factors and to relieve the common problem of headaches in 31 32 33 12 healthcare workers. 34 35 13 Our study additionally confirmed the negative effects of other types of chronic pain 36

37 http://bmjopen.bmj.com/ 38 14 on sleep, suggesting that chronic pain in different parts of the body increases the 39 40 15 likelihood of sleep problems. A study by Andersen et al. on healthcare workers also 41 42 43 16 indicated that chronic musculoskeletal pain in one body region increases the risk of 44

45 24 on September 27, 2021 by guest. Protected copyright. 46 17 developing chronic pain in other pain-free body regions . Therefore, when 47 48 18 healthcare workers complain of chronic pain, they should be diagnosed and treated in 49 50 51 19 a timely manner, to not only reduce the risk of sleeping problems, but also avoid 52 53 20 causing chronic pain in other parts. 54 55 56 21 Our finding that night-shift work predicted a greater risk of sleep problems is 57 58 9 34 59 22 consistent with the results of several previous studies . In hospitals, inpatient 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 services usually require round-the-clock services, which often mean healthcare 5 6 7 2 workers have to work extended shifts (especially night shifts) and long hours, which 8 9 3 deprive them of sleep. Moreover, night work leads to misalignments between internal 10 11 12 4 circadian rhythms and work-rest schedules, again resulting in fewer sleeping hours 46. 13 14 5 As such, finding effective ways to minimize the negative effects of shift work on 15 16 17 6 health workers’ sleep is essential. Previous research suggests that timed bright lights 18 For peer review only 19 47 48 20 7 and light-blocking goggles can promote circadian adaptation to shift work . 21 22 8 However, more reasonable night work schedules for healthcare workers may be a 23 24 25 9 better solution to improve their sleep problems. Our data suggests that the number of 26 27 10 night shifts per month for healthcare workers should be reduced at reasonable level. 28 29 30 11 Meanwhile, according to some previous studies, we also suggest that it is best to 31 32 49 33 12 allow for sufficient duration of time off between shifts and set a nap time to 34 35 13 alleviate the high levels of sleepiness experienced during night work 49 50. 36

37 http://bmjopen.bmj.com/ 38 14 The present study also revealed that long working hours were a critical risk factor 39 40 15 associated with sleep problems. A cross-sectional study in the United States reported 41 42 43 16 that compensated working time was consistently associated with sleep duration, such 44

45 51 on September 27, 2021 by guest. Protected copyright. 46 17 that work time increased as sleep time decreased . In addition, the working hours of 47 48 18 healthcare workers are obviously too long. According to the latest survey data 49 50 51 19 collected in 2017, healthcare workers in China’s tertiary hospitals work an average of 52 53 20 51.05 h per week 52, which is similar to the hours that were reported by participants in 54 55 56 21 our study. Hale et al. demonstrated that people working >50 h per week were more 57 58 53 59 22 likely to be short sleepers . Many countries have clear rules limiting the maximum 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 working hours of healthcare workers. For example, the European Union specified a 5 6 7 2 maximum of 48 h of work per week and established rest periods for healthcare 8 9 3 workers in 2009 54. Although China’s labor law specifies a maximum working time, it 10 11 12 4 is not applicable to the profession of healthcare workers 55. Currently, there are no 13 14 5 special policies on the labor rights of healthcare workers 55. 15 16 17 6 Our findings confirmed that healthy behaviors play an important role in sleep 18 For peer review only 19 20 7 problems. Indeed, our results showed that more than three-quarters of healthcare 21 22 8 workers were unable to maintain a regular diet, and an irregular diet was significantly 23 24 25 9 associated with sleep problems. Many healthcare workers do not have enough time to 26 27 10 eat while working, and thus they order fast food from delivery services or simply 28 29 30 11 delay their meal. However, research has revealed that short sleep durations are 31 32 33 12 associated with higher energy intake, mostly due to increased consumption of 34 35 13 saturated fat 56. We recommend that healthcare workers should eat regular diet and to 36

37 http://bmjopen.bmj.com/ 38 14 increase the intake of micronutrients such as tryptophan and vitamin Bs in their diets 39 40 15 to improve sleep problems 57. In addition, we also observed that participants who did 41 42 43 16 not engage in regular exercise were more likely to experience sleep problems, which 44

45 58 on September 27, 2021 by guest. Protected copyright. 46 17 is consistent with the findings of previous research . A randomized controlled trial 47 48 18 found that improvement of sleep quality after exercise intervention were evident and 49 50 51 19 physical activity is beneficial in terms of quality and duration of sleep 57. For adults 52 53 20 aged 18–64 years, the World Health Organization recommends at least 150 min of 54 55 56 21 moderate-intensity aerobic physical activity throughout the week, or at least 75 min of 57 58 59 59 22 vigorous-intensity aerobic physical activity throughout the week . 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 A study funded by the Chinese Medical Doctor Association showed that 46 cases of 5 6 7 2 karoshi, or “death by overwork,” were reported among Chinese healthcare workers 8 9 3 between 2013 and 2015, and most cases occurred in healthcare workers aged 30–39 10 11 12 4 years 60. Although healthcare workers in China’s top-three hospitals treat many 13 14 5 patients, few of them recognize their own health. As was found here, healthcare 15 16 17 6 workers generally have health-related problems, such as lack of sleep, lack of physical 18 For peer review only 19 20 7 exercise, and an irregular diet. Studies have shown that continuous exposure to these 21 22 8 factors may increase the risk of chronic diseases among healthcare workers 10 61 62. 23 24 25 9 The phenomenon of healthcare workers not paying proper attention to their own 26 27 10 health should be recognized by the government and society. 28 29 30 11 Overall, existing evidence indicates that it is important for healthcare workers to 31 32 33 12 increase their sleep durations and reduce their workloads. The effects that chronic 34 35 13 pain, especially headaches, has on the healthcare workers’ sleep problems are a 36

37 http://bmjopen.bmj.com/ 38 14 special area of concern that requires attention. Furthermore, ensuring workers are 39 40 15 engaging in healthy behaviors is equally important. Based on our findings, we make 41 42 43 16 the following recommendations: (1) hospitals should ensure healthcare workers’ 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 chronic pain in different parts of body be diagnosed and treated in a timely manner; 47 48 18 (2) healthcare workers should beware of their own health issues, and pay more 49 50 51 19 attention to cultivating healthy behaviors, such as eating a regular diet, and getting 52 53 20 regular physical examinations, etc.; (3) more rational working hours and night-shift 54 55 56 21 work systems should be established; and (4) health administrative departments should 57 58 59 22 expand the scope of labor laws and formulate special regulations for the labor rights 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 of healthcare workers. 5 6 7 2 Our study has some limitations of note. First, information regarding the exposure and 8 9 3 outcome was obtained via a self-report questionnaire, which may introduce report bias 10 11 12 4 in these variables. In particular, we determined sleep problems via subjective 13 14 5 evaluations of the participants. Future research should incorporate validated scales 15 16 17 6 and clinical diagnoses to support the results obtained via self-report. Second, all 18 For peer review only 19 20 7 participants were from Heilongjiang Province, China, and thus the study findings may 21 22 8 not be generalizable to other populations. Moreover, the sample did not include 23 24 25 9 nursing staff, and hence the results cannot be used to characterize all healthcare 26 27 10 workers. Third, as this was a cross-sectional study, we cannot establish causal 28 29 30 11 relationships based on the results. Fourth, only one subpopulation of healthcare 31 32 33 12 workers was examined to test for effect modification. Finally, survey weights were 34 35 13 not considered in this study, and the results were based on an unweighted analysis, the 36

37 http://bmjopen.bmj.com/ 38 14 odds ratio of which might be smaller than that of considering weight 63. Therefore, 39 40 15 sample weighting should be conducted in future analyses. 41 42 43 16 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 Conclusions 47 48 18 In summary, our findings suggest that nearly half of the healthcare workers in tertiary 49 50 51 19 hospitals from Heilongjiang Province in China have sleep problems. Chronic pain 52 53 20 (headaches, knee pain, neck and shoulder pain, and lumbar and back pain), long 54 55 56 21 working hours, and frequent night shifts were positively associated with sleep 57 58 59 22 problems among healthcare workers. Of these, headaches were the most prominent 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 risk factor for sleep problems in this population. However, a regular diet and daily 5 6 7 2 exercise may reduce the likelihood of sleep problems. Most findings were not altered 8 9 3 when analyzing only participants with short sleep durations. These results indicate 10 11 12 4 that to improve the sleep problems of healthcare workers, both medical institutions 13 14 5 and health policy makers should work hand in hand to provide healthcare workers 15 16 17 6 with reasonable work schedules, while the healthcare workers themselves should 18 For peer review only 19 20 7 focus on their chronic pain problems and engaging in regular diet and exercise. 21 22 8 23 24 25 9 Acknowledgements 26 27 10 The authors are very grateful to Miaomiao Zhao, Ding Ding, Linghan Shan for 28 29 30 11 providing the suggestion to draft the manuscript. We also thank all the participating 31 32 33 12 organizations and participants. 34 35 13 36

37 http://bmjopen.bmj.com/ 38 14 Footnotes 39 40 15 Funding 41 42 43 16 This study was supported by the CMB Distinguished Professorship Awards (Grant 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 No. G16916400). 47 48 18 49 50 51 19 Competing interests 52 53 20 The authors declare that they have no competing interests. 54 55 56 21 57 58 59 22 Contributors 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 HL conducted calculations, analyzed results and drafted the manuscript. QHW and 5 6 7 2 YHH were responsible for the overall design of the research, organized and conducted 8 9 3 the survey. MLJ, NN and LL designed the analyses framework. ZK and LJG revised 10 11 12 4 the paper. JJL, MC, XT and TZ assisted with the literature review and data collection. 13 14 5 All authors approval of the current version of this manuscript for publications. 15 16 17 6 18 For peer review only 19 20 7 Patient consent 21 22 8 Patient consent was obtained for publication 23 24 25 9 26 27 10 Data sharing statement 28 29 30 11 No additional data are available. 31 32 33 12 34 35 13 References 36

37 14 1. Koyanagi A, Stickley A. The Association between Sleep Problems and Psychotic Symptoms in the http://bmjopen.bmj.com/ 38 15 General Population: A Global Perspective. Sleep 2015;38(12):1875-85. doi: 39 40 16 10.5665/sleep.5232 [published Online First: 2015/06/19] 41 17 2. Owens JA, Mindell JA. A Clinical Guide to Pediatric Sleep. Journal of the American Academy of Child 42 18 & Adolescent Psychiatry 2005;44(7):720–21. 43 44 19 3. Chinese Medical Association. China Sleep Quality Index announced 2014. 2014. 4. Sun W, Fu J, Chang Y, et al. Epidemiological study on risk factors for anxiety disorder among

45 20 on September 27, 2021 by guest. Protected copyright. 46 21 Chinese doctors. Journal of Occupational Health 2012;54(1):1-8. 47 48 22 5. He Tian, Yancheng Di, Song J. Investigation of sleep quality and influncing factors of medical 49 23 workers in a grade -A tertiary hospitals. Chinese Journal of Social Medicine 2015;32(1):48-51. 50 24 6. Yang X, Wang R, Liu Z, et al. Relationship between Sleep Quality and Mental Health of Physicians. 51 25 China Journal of Health Psychology 2007;15(7):654-56. 52 53 26 7. Hayashino Y, Fukuhara S, Suzukamo Y, et al. Relation between sleep quality and quantity, quality of 54 27 life, and risk of developing diabetes in healthy workers in Japan: the High-risk and Population 55 28 Strategy for Occupational Health Promotion (HIPOP-OHP) Study. BMC Public Health 56 57 29 2007;7:129. doi: 10.1186/1471-2458-7-129 [published Online First: 2007/06/29] 58 30 8. Sleep deprivation, sleep disorders, fatigue, stress and the risk of occupational injuries and errors 59 31 among health-care workers: a multi-approach epidemiological study. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 43

1 2 3 1 9. Ghalichi L, Pournik O, M G. Sleep quality among health care workers. Arch Iran Med BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 2013;16(2):100-03. 6 3 10. Buxton OM, Marcelli E. Short and long sleep are positively associated with obesity, diabetes, 7 4 hypertension, and cardiovascular disease among adults in the United States. Social Science & 8 9 5 Medicine 2010;71(5):1027-36. 10 6 11. Irwin MR. Why sleep is important for health: a psychoneuroimmunology perspective. Annu Rev 11 7 Psychol 2015;66:143-72. doi: 10.1146/annurev-psych-010213-115205 [published Online 12 8 First: 2014/07/26] 13 14 9 12. Gottlieb DJ, Redline S, Nieto FJ, et al. Association of usual sleep duration with hypertension: the 15 10 Sleep Heart Health Study. Sleep 2006;29(8):1009-14. 16 11 13. Kling RN, McLeod CB, Koehoorn M. Sleep problems and workplace injuries in Canada. Sleep 17 18 12 2010;33(5):611-8.For doi: peer 10.1093/sleep/33.5.611 review [published only Online First: 2010/05/18] 19 13 14. Eddy R. Sleep deprivation among physicians. British Columbia Medical Journal 2005(4):176-80. 20 14 15. Ayas NT, Barger LK, Cade BE, et al. Extended work duration and the risk of self-reported 21 22 15 percutaneous injuries in interns. JAMA 2006;296(9):1055-62. doi: 10.1001/jama.296.9.1055 23 16 [published Online First: 2006/09/07] 24 17 16. Chaudhury PK, Warnock GL, Whalen TV, et al. CAGS and ACS evidence based reviews in surgery. 25 18 Risks of complications by attending physicians after performing nighttime procedures. Can J 26 27 19 Surg 2012;55(5):337-9. doi: 10.1503/cjs.018212 [published Online First: 2012/09/21] 28 20 17. Mycyk MB, McDaniel MR, Fotis MA, et al. Hospitalwide adverse drug events before and after 29 21 limiting weekly work hours of medical residents to 80. Am J Health Syst Pharm 30 31 22 2005;62(15):1592-5. doi: 10.2146/ajhp040527 [published Online First: 2005/07/21] 32 23 18. Lockley SW, Landrigan CP, Barger LK, et al. When policy meets physiology: the challenge of 33 24 reducing resident work hours. Clin Orthop Relat Res 2006;449:116-27. doi: 34 35 25 10.1097/01.blo.0000224057.32367.84 [published Online First: 2006/06/14] 36 26 19. Department of Economic and Social Affairs Population Division. World population ageing 2015.

37 27 United Nations. 2015; 135-154.18. Chen W, Zheng R, Baade PD, et al. http://bmjopen.bmj.com/ 38 28 20. Cancer statistics in China, 2015. CA Cancer J Clin 2016;66(2):115-32. doi: 10.3322/caac.21338 39 40 29 [published Online First: 2016/01/26]20. Chen W, Zheng R, Baade PD, et al. Cancer statistics in 41 30 China, 2015. CA Cancer J Clin 2016;66(2):115-32. doi: 10.3322/caac.21338 [published Online 42 31 First: 2016/01/26] 43 44 32 21. Moran A, Gu D, Zhao D, et al. Future cardiovascular disease in china: markov model and risk factor scenario projections from the coronary heart disease policy model-china. Circ Cardiovasc

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 Qual Outcomes 2010;3(3):243-52. doi: 10.1161/CIRCOUTCOMES.109.910711 [published 47 48 35 Online First: 2010/05/06] 49 36 22. Kobayashi T, Suzuki E, Takao S, et al. Long working hours and metabolic syndrome among 50 37 Japanese men: a cross-sectional study. Bmc Public Health 2012;12(1):395. 51 38 23. Kim MS, Kim T, Lee D, et al. Mental disorders among workers in the healthcare industry: 2014 52 53 39 national health insurance data. Ann Occup Environ Med 2018;30:31. doi: 10.1186/s40557- 54 40 018-0244-x [published Online First: 2018/05/15] 55 41 24. Andersen LL, Clausen T, Carneiro IG, et al. Spreading of chronic pain between body regions: 56 57 42 prospective cohort study among health care workers. Eur J Pain 2012;16(10):1437-43. doi: 58 43 10.1002/j.1532-2149.2012.00143.x [published Online First: 2012/03/31] 59 44 25. Jank R, Gallee A, Boeckle M, et al. Chronic Pain and Sleep Disorders in Primary Care. Pain Res Treat 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 43 BMJ Open

1 2 3 1 2017;2017:9081802. doi: 10.1155/2017/9081802 [published Online First: 2018/02/08] BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 26. Stonge MP, Mikic A, Pietrolungo CE. Effects of Diet on Sleep Quality. Advances in Nutrition 6 3 2016;7(5):938. 7 4 27. Buysse DJ. Sleep health: can we define it? Does it matter? Sleep 2014;37(1):9. 8 9 5 28. Knutson KL, Phelan J, Paskow MJ, et al. The National Sleep Foundation's Sleep Health Index. Sleep 10 6 Health 2017;3(4):234. 11 7 29. Consensus Conference P, Watson NF, Badr MS, et al. Joint Consensus Statement of the American 12 8 Academy of Sleep Medicine and Sleep Research Society on the Recommended Amount of 13 14 9 Sleep for a Healthy Adult: Methodology and Discussion. Sleep 2015;38(8):1161-83. doi: 15 10 10.5665/sleep.4886 [published Online First: 2015/07/22] 16 11 30. Seyedmehdi SM, Rahnama N, Yazdanparast T, et al. Prevalence of snoring and the risk of sleep 17 18 12 apnea in hospitalFor staff. peer Work 2016;55(4):765-72. review doi: 10.3233/WOR-162460 only [published 19 13 Online First: 2017/01/07] 20 14 31. National Bureau of Statistics. China health statistical yearbook 2017. Beijing Union Medical 21 University Press. 2017; 118:138. 22 15 23 16 32. Chinese Medical Doctor Association. Chinese doctors practicing white paper 2015. 24 17 http://www.cmda.net/zlwqgzdt/596.jhtml [Accessed 14 Apr 2018]. 25 18 33. Jafari Roodbandi A, Choobineh A, Daneshvar S. Relationship between circadian rhythm amplitude 26 27 19 and stability with sleep quality and sleepiness among shift nurses and health care workers. 28 20 Int J Occup Saf Ergon 2015;21(3):312-7. doi: 10.1080/10803548.2015.1081770 [published 29 21 Online First: 2015/09/29] 30 31 22 34. ALEMOHAMMAD ZB, TAGHAVI SME, SHARIFIAN A. Sleep Quality and Restless Legs Syndrome 32 23 among Health-care Workers: Shift Workers and Non-shift Workers. Journal of Sleep Sciences 33 24 2016;1(2):74-79. 34 35 25 35. Kolo ES, Ahmed AO, Hamisu A, et al. Sleep health of healthcare workers in Kano, Nigeria. Niger J 36 26 Clin Pract 2017;20(4):479-83. doi: 10.4103/1119-3077.204378 [published Online First:

37 27 2017/04/14] http://bmjopen.bmj.com/ 38 28 36. National Health and Family Planning Commission. China health statistical yearbook 2015. Beijing: 39 40 29 China Union Medical University Press, 2015; 11. 41 30 37. Stranges S, Tigbe W, Gomez-Olive FX, et al. Sleep problems: an emerging global epidemic? 42 31 Findings from the INDEPTH WHO-SAGE study among more than 40,000 older adults from 8 43 44 32 countries across Africa and Asia. Sleep 2012;35(8):1173-81. doi: 10.5665/sleep.2012 [published Online First: 2012/08/02]

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 38. Merskey H. International Association for the Study of Pain: Classification of chronic pain. 47 48 35 Descriptions of chronic pain syndromes and definitions of pain terms. Pain 1986; 3(2):S1- 49 36 S226. 50 37 39. Chen W, Zhou F, Hall BJ, et al. Is there a relationship between geographic distance and uptake of 51 38 HIV testing services? A representative population-based study of Chinese adults in 52 53 39 Guangzhou, China. PLoS One 2017;12(7):e0180801. doi: 10.1371/journal.pone.0180801 54 40 [published Online First: 2017/07/21] 55 41 40. Wang J, Xie H, Fisher JH. Multilevel models: applications using SAS2009. 56 57 42 41. Ju H, Pan XL, Li A. Multi-level modelling of the factors that influence the participation of disabled 58 43 rural individuals in social medical insurance in China. Bmc Health Services Research 59 44 2013;13(1):58-58. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 43

1 2 3 1 42. Ertel KA, Berkman LF, Buxton OM. Socioeconomic status, occupational characteristics, and sleep BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 duration in African/Caribbean immigrants and US White health care workers. Sleep 6 3 2011;34(4):509-18. [published Online First: 2011/04/05] 7 4 43. Burns RD. Energy balance-related factors associating with adolescent weight loss intent: evidence 8 9 5 from the 2017 National Youth Risk Behavior Survey. BMC Public Health 2019;19(1):1206. doi: 10 6 10.1186/s12889-019-7565-8 [published Online First: 2019/09/04] 11 7 44. Onwuekwe I, Onyeka T, Aguwa E, et al. Headache prevalence and its characterization amongst 12 8 hospital workers in Enugu, South East Nigeria. Head & face medicine 2014;10(1):48. 13 14 9 45. Holland PR. Headache and sleep: shared pathophysiological mechanisms. Cephalalgia 15 10 2014;34(10):725-44. doi: 10.1177/0333102414541687 [published Online First: 2014/07/24] 16 11 46. Zhang Y, Punnett L, Mcenany GP, et al. Contributing influences of work environment on sleep 17 18 12 quantity andFor quality peerof nursing assistants review in long-term care only facilities: A cross-sectional study. 19 13 Geriatric Nursing 2015;37(1):13-18. 20 14 47. Burgess HJ, Sharkey KM, Eastman CI. Bright light, dark and melatonin can promote circadian 21 22 15 adaptation in night shift workers. Sleep Medicine Reviews 2002;6(5):407-20. 23 16 48. Neil-Sztramko SE, Pahwa M, Demers PA, et al. Health-related interventions among night shift 24 17 workers: a critical review of the literature. Scandinavian Journal of Work Environment & 25 18 Health 2014;40(6):543-56. 26 27 19 49. Li J, Bidlingmaier M, Petru R, et al. Impact of shift work on the diurnal cortisol rhythm: a one-year 28 20 longitudinal study in junior physicians. J Occup Med Toxicol 2018;13:23. doi: 10.1186/s12995- 29 21 018-0204-y [published Online First: 2018/08/21] 30 31 22 50. Rebecca SC, Howard SK, Mac DT, et al. Improving alertness and performance in emergency 32 23 department physicians and nurses: the use of planned naps. Annals of Emergency Medicine 33 24 2006;48(5):596-604.e3. 34 35 25 51. Basner M, Fomberstein KM, Razavi FM, et al. American Time Use Survey: Sleep Time and Its 36 26 Relationship to Waking Activities. Sleep 2007;30(9):1085-95.

37 27 52. Chinese Medical Doctor Association. Chinese doctors practicing white paper 2017. http://bmjopen.bmj.com/ 38 28 http://www.cmda.net/u/cms/www/201807/06181247ffex.pdf. [Accessed 20 May 2018] 39 40 29 53. Hale L. Who has time to sleep? J Public Health 2005;27(2):205-11. 41 30 54. Lancet T. Doctors’ training and the European Working Time Directive. Lancet 42 31 2010;375(9732):2121. 43 44 32 55. Ren Y. Study on Labor Rights Guarantee of Medical Staff in China. Legal System and Society 2017(4):293-94.

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 56. Patel SR, Hu FB. Short sleep duration and weight gain: a systematic review. Obesity 47 48 35 2008;16(3):643. 49 36 57. Tan X. Effects of exercise and diet interventions on obesity-related sleep disorders in men: study 50 37 protocol for a randomized controlled trial. Trials 2013;14(1):1-9. 51 38 58. Han Y, Yuan Y, Zhang L, et al. Sleep disorder status of nurses in general hospitals and its 52 53 39 influencing factors. Psychiatria Danubina 2016;28(2):176-83. 54 40 59. Organization WH. Global recommendations on physical activity for health. 2010 55 41 60. Shan HP, Yang XH, Zhan XL, et al. Overwork is a silent killer of Chinese doctors: a review of Karoshi 56 57 42 in China 2013-2015. Public Health 2017;147:98. 58 43 61. Thomas SJ, Calhoun D. Sleep, insomnia, and hypertension: current findings and future directions. J 59 44 Am Soc Hypertens 2017;11(2):122-29. doi: 10.1016/j.jash.2016.11.008 [published Online 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 43 BMJ Open

1 2 3 1 First: 2017/01/23] BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 62. Yu H, Xie LF, Chen K, et al. Initiating Characteristics of Early-onset Type 2 Diabetes Mellitus in 6 3 Chinese Patients. Chin Med J (Engl) 2016;129(7):778-84. doi: 10.4103/0366-6999.178959 7 4 [published Online First: 2016/03/22] 8 9 5 63. Bethany AB JD. Alternatives for Analysis of Complex Sample Surveys: A Comparison of SAS, 10 6 SUDAAN and AM software. SAS Global Forum 2007:2659-66. 11 12 7 13 14 15 8 Figure legend 16 17 9 Figure 1. Subjective sleep duration (hours per night) 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 43 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 Figure 1. Subjective sleep duration (hours per night) 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 27, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 43 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 Additional file 1. Questionnaire 5 6 7 8 ID number：______9 10 [1]Name：______11 ： 12 [2]Age ______13 [3]Gender：_____ ①Male ②Female 14 15 [4]Hospital：______16 For peer review only 17 [5]Department： 18 19 [5.1]Clinical department：______(If you are not a healthcare worker in the clinical 20 21 department, please fill in 0.) 22 23 ①Internal Medicine ②Surgery Department 24 25 ③Obstetricians and Gynecologists ④Pediatrics Department 26 27 ⑤Ophthalmology Department ⑥Stomatology Department 28 29 ⑦Otolaryngology Department ⑧Dermatology Department 30 31 ⑨Anesthesiology Department ⑩Emergency Department 32 ⑪Traditional Chinese Medicine Department ⑫Infectious Diseases Department 33 http://bmjopen.bmj.com/ 34 35 ⑬Plastic Surgery Department ⑭ICU ⑮Others ______36 [5.2] Medical technology department：______(If you are not a healthcare worker in 37 38 the medical technology department, please fill in 0.) 39 40 ①Pharmacy Department ②Clinical Lab ③Radiology Department 41 on September 27, 2021 by guest. Protected copyright. 42 43 ④Pathology Department ⑤Nuclear Medicine Department ⑥Nutritional Department 44 45 ⑦Others______46 47 [5.3] Administrative department：______(If you are not a healthcare worker in the 48 49 medical technology department, please fill in 0.) 50 51 ①Personnel Department ②Preventive care department 52 53 ③Medical Service Management Office ④Finance Department 54 55 ⑤Administration Office ⑥Others______56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 43 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 [6]Educational background：______5 6 ①Doctor ②Master ③Bachelor ④Junior college 7 8 ⑤High school or technical secondary school ⑥Others______9 10 [7]Title：______11 ① ② ③ ④ 12 senior title intermediate job title junior titles No title 13 ⑤Others______14 15 [8] What is your monthly salary now? ______yuan 16 For peer review only 17 [9] How much time do you work on average per week? ______Hours 18 19 [10] What's your average monthly night shift (over half an hour and above)？ 20 21 ______time(s)/per month 22 23 [11] How much time do you spend on your daily one-way commute? ______24 25 ①<30 minutes ②30-59 minutes ③60-89 minutes ④>=90 minutes 26 27 [12] Was your diet regular at work? ______28 29 ①Yes ②No 30 31 [13] Did you defecate regularly at work? ______32 ①Yes ②No 33 http://bmjopen.bmj.com/ 34 35 [14] Do you have time for physical exercise every day? ______36 ①Yes ②No 37 38 [15] How many hours do you usually sleep a day in the past month？______hour(s) 39 40 [16] Overall, in the last 30 days, how much of a problem did you have with sleeping,

41 on September 27, 2021 by guest. Protected copyright. 42 such as falling asleep, waking up frequently during the night, or waking up too early in 43 44 the morning? ______45 46 ①None ②Mild ③Moderate ④Severe ⑤Extreme 47 48 [17]Do you have experienced chronic pain that had not improved in the past 30 days in 49 50 any of the following regions？ 51 52 [Note: the chronic pain was defined as any continuous or persistent intermittent pain 53 54 experienced for a period longer than 3 months] 55 56 [17.1] Headaches ______①Yes ②No 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 43 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 [17.2] Neck and shoulder pain ______①Yes ②No 5 6 [17.3] Lumbar and back pain ______①Yes ②No 7 8 [17.4] Knee pain ______①Yes ②No 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 27, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 Additional file 2. Individual-level variables and coding 5 6 7 8 9 Table S1. Individual-level variables and coding for regression modeling. 10 11 Variables Characteristic Coding 12 Sex 13 Women 0 14 15 Men 1 16 Age (years) 17 < 30 1 18 For peer review30–40 only 2 19 20 > 40 3 21 Educational background 22 Bachelor degree and below 0 23 24 Master's degree and above 1 25 Job title 26 Junior titles and others 1 27 28 Intermediate title 2 29 Senior title 3 30 Monthly salary income 31 ≤ 3,000 yuan 0 32 33 > 3,000 yuan 1 34 One-way commute time 35 < 30 minutes 0 36 ≥ 30 minutes 1

37 http://bmjopen.bmj.com/ 38 Regular diet at work 39 No 0 40 41 Yes 1 42 Work hours per week 43 (hours) 44 < 55 0

45 on September 27, 2021 by guest. Protected copyright. 46 ≥ 55 1 47 Night shifts per month 48 (times) 49 50 > 5 0 51 ≤ 5 1 52 Daily exercise 53 54 No 0 55 Yes 1 56 Headaches 57 No 0 58 59 Yes 1 60 Neck and shoulder pain

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 43 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 No 0 5 Yes 1 6 Lumbar and back pain 7 No 0 8 9 Yes 1 10 Knee pain 11 No 0 12 13 Yes 1 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 43

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 Additional file 3. The names of Ethics Committees 5 6 7 Coding The names of Ethics Committees 8 9 1 The Ethics Committee of Daqing Longnan Hospital 10 2 The Ethics Committee of General Hospital of Daqing Oil Field 11 12 3 The Ethics Committee of Daqing City People's Hospital 13 4 The Ethics Committee of Harbin 242 Hospital 14 15 5 The Ethics Committee of The Fifth Hospital of Harbin 16 6 The Ethics Committee of Harbin The First Hospital 17 7 The Ethics Committee of Harbin First Specialist Hospital 18 For peer review only 19 8 The Ethics Committee of Harbin Red Cross Central Hospital 20 9 The Ethics Committee of The Third Affiliated Hospital of Harbin Medical University 21 22 10 The Ethics Committee of The Fourth Affiliated Hospital of Harbin Medical University 23 11 The Ethics Committee of The First Affiliated Hospital of Harbin Medical University 24 25 12 The Ethics Committee of Heilongjiang Land Reclamation Bureau Hongxinglong Bureau Central Hospital 26 13 The Ethics Committee of The Fourth Hospital of Heilongjiang Province 27 28 14 The Ethics Committee of Heilongjiang Province Land Reclamation Headquarters General Hospital 29 15 The Ethics Committee of Heilongjiang Red Cross Sengong General Hospital 30 31 16 The Ethics Committee of Heilongjiang Provincial Hospital of Traditional Chinese medicine 32 17 The Ethics Committee of The Second Affiliated Hospital of Heilongjiang University of Chinese Medicine 33 34 18 The Ethics Committee of The First Affiliated Hospital of Heilongjiang University of Chinese Medicine 35 19 The Ethics Committee of Jixi Mining Group General Hospital 36 37 20 The Ethics Committee of Jixi People's Hospital http://bmjopen.bmj.com/ 38 21 The Ethics Committee of Jixi Hospital of Traditional Chinese Medicine 39 40 22 The Ethics Committee of The Second Affiliated Hospital of Jiamusi University 41 23 The Ethics Committee of Jiamusi City Tuberculosis Hospital 42 43 24 The Ethics Committee of Jiamusi Center Hospital 44 25 The Ethics Committee of Hongqi Hospital Affiliated to Mudanjiang Medical University 45 on September 27, 2021 by guest. Protected copyright. 46 26 The Ethics Committee of Qiqihar Third Hospital 47 27 The Ethics Committee of the Second Affiliated Hospital of Qiqihar Medical College 48 49 28 The Ethics Committee of Qiqihar Traditional Chinese Medicine Hospital 50 29 The Ethics Committee of the Third Affiliated Hospital of Qiqihar Medical University 51 52 30 The Ethics Committee of Heilongjiang Provincial Hospital 53 31 The Ethics Committee of Shuangyashan Coal General Hospital 54 55 32 The Ethics Committee of The Shuangyashan Mining Bureau General Hospital 56 33 The Ethics Committee of Yichun Forestry Central Hospital 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Page 41 of 43 BMJ Open

1 2 STROBE 2007 (v4) Statement—Checklist of items that should be included in reports of cross-sectional studies 3 4 5 Item Section/Topic Recommendation Reported on page # 6 # 7 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract Page 4, line 11 8 9 (b) Provide in the abstract an informative and balanced summary of what was done and what was found Page 4, line 18-21; 10 Page 4, line 22; 11 Page 5, line 1- 5. 12 For peer review only 13 Introduction 14 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported Page 7-9; 15 Page 8, line 1-6 16 17 Objectives 3 State specific objectives, including any prespecified hypotheses http://bmjopen.bmj.com/ Page 10, line 3-12 18 Methods 19 20 Study design 4 Present key elements of study design early in the paper Page 8, line 16-22; 21 Page 9, line 1-19; 22 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data Page 10, line 16-23; 23 collection Page 11, line 1-10 24 25 Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of participants on September 27, 2021 by guest. Protected copyright. Page 10, line 8-10 26 27 28 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if Page 11, line 12-22; 29 applicable Page 12; 30 Page 13, line 3-12; 31 32 Page 14, line 10-12 33 Data sources/ 8* For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe Page 11, line 12-22; 34 measurement comparability of assessment methods if there is more than one group Page 12; 35 Page 13, line 3-12; 36 37 Page 14, line 10-12 38 Bias 9 Describe any efforts to address potential sources of bias No 39 Study size 10 Explain how the study size was arrived at No 40 41 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and Page 11, line 12-22; 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open Page 42 of 43

1 2 why Page 12; 3 Page 13, line 3-12 4 5 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding Page 13, line 14-22; 6 Page 14, line 1-16; 7 (b) Describe any methods used to examine subgroups and interactions No 8 9 (c) Explain how missing data were addressed Page 11, line 5-10 10 (d) If applicable, describe analytical methods taking account of sampling strategy Page 10, line 22-23; 11 Page 11, line 1-6 12 For peer review only 13 (e) Describe any sensitivity analyses No 14 Results 15 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, examined for eligibility, Page 10, line 16-23; 16 17 confirmed eligible, included in the study, completing follow-up, and analysed http://bmjopen.bmj.com/ Page 11, line 1-10 18 (b) Give reasons for non-participation at each stage No 19 (c) Consider use of a flow diagram No 20 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential Page 15, line 13-18 21 22 confounders 23 (b) Indicate number of participants with missing data for each variable of interest No 24 Outcome data 15* Report numbers of outcome events or summary measures Page 15, line 13-18 25 on September 27, 2021 by guest. Protected copyright. 26 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence Page 17, line 9-14; 27 interval). Make clear which confounders were adjusted for and why they were included Page 18, line 1-25 28 (b) Report category boundaries when continuous variables were categorized Page 11, line 12-22; 29 Page 12; 30 31 Page 13, line 3-12; 32 Page 14, line 10-12 33 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period No 34 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses Page 19, line 3-5; 35 36 Page 20, line 1-3 37 Discussion 38 Page 22, line 3-9 39 Key results 18 Summarise key results with reference to study objectives 40 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and Page 27, line 2-15 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Page 43 of 43 BMJ Open

1 2 magnitude of any potential bias 3 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from Page 27, line 18-22; 4 5 similar studies, and other relevant evidence Page 28, line 1-7; 6 Generalisability 21 Discuss the generalisability (external validity) of the study results Page 27, line 2-10 7 8 Other information 9 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on Page 28, line 16 10 which the present article is based 11 12 *Give information separately for cases and controlsFor in case-control peer studies and, if applicable,review for exposed and unexposedonly groups in cohort and cross-sectional studies. 13 14 15 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE 16 checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 17 http://bmjopen.bmj.com/ 18 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org. 19 20 21 22 23 24 25 on September 27, 2021 by guest. Protected copyright. 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Sleep problems of healthcare workers in tertiary hospital and influencing factors identified through a multi-level analysis: a cross-sectional study in China. ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-032239.R2

Article Type: Original research

Date Submitted by the 02-Dec-2019 Author:

Complete List of Authors: liu, huan; Harbin Medical University, Liu, jing; Harbin Medical University School of Public Health, Department of Social Medicine chen, mingxi; Harbin Medical University, Social Medicine tan, xiao; Harbin Medical University, Social Medicine zheng, tong; Harbin Medical University, Social Medicine Kang, Zheng; Harbin Medical University, Social Medicine; Gao, Lijun; Harbin Medical University, Social Medicine Jiao, Mingli; Harbin Medical University, Health policy Ning, Ning; Harbin Medical University, Social Medicine Liang, Libo; School of Health Management, Harbin Medical University, Social Medicine; Wu, Qunhong; Harbin Medical University, Social Medicine http://bmjopen.bmj.com/ Hao, Yanhua; Harbin Medical University, Social Medicine

Primary Subject Public health Heading:

Secondary Subject Heading: Health policy, Occupational and environmental medicine, Public health

Healthcare worker, Tertiary hospitals, Sleep problems, Influencing Keywords: factors on September 27, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Title Page 5 6 7 2 8 9 3 Title 10 11 4 Sleep problems of healthcare workers in tertiary hospital and influencing factors 12 5 identified through a multi-level analysis: a cross-sectional study in China. 13 14 6 15 16 7 Author names, affiliations and email addresses 17 1 1 1 1 1 1 18 8 Huan Liu , JingjingFor Liu ,peer Mingxi Chen review, Xiao Tan , Tong only Zheng , Zheng Kang , 19 9 Lijun Gao1, Mingli Jiao2, Ning Ning1, Libo Liang1, Qunhong Wu1*, Yanhua Hao1*. 20 21 10 22 23 1 24 11 Department of Social Medicine, Health Management College, Harbin Medical 25 26 12 University, Harbin, Heilongjiang Province, China. 27 28 2 29 13 Department of Health Policy, Health Management College, Harbin Medical University, 30 31 14 Harbin, Heilongjiang Province, China. 32 33 15 34 35 16 Huan Liu, Department of Social Medicine, Health Management College, Harbin 36 17 Medical University, Harbin, Heilongjiang Province, China. Email:

37 http://bmjopen.bmj.com/ 38 39 18 [email protected] 40 41 19 Jingjing Liu, Department of Social Medicine, Health Management College, Harbin 42 43 20 Medical University, Harbin, Heilongjiang Province, China. Email: 44 21 [email protected]

45 on September 27, 2021 by guest. Protected copyright. 46 47 22 Mingxi Chen, Department of Social Medicine, Health Management College, Harbin 48 49 23 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 50 Xiao Tan, Department of Social Medicine, Health Management College, Harbin 51 24 52 25 Medical University, Harbin, Heilongjiang Province, China. Email: 53 54 26 [email protected] 55 56 27 Tong Zheng, Department of Social Medicine, Health Management College, Harbin 57 58 28 Medical University, Harbin, Heilongjiang Province, China. Email: 59 60 29 [email protected]

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Zheng Kang, Department of Social Medicine, Health Management College, Harbin 5 6 2 Medical University, Harbin, Heilongjiang Province, China. Email: 7 8 3 [email protected] 9 10 4 Lijun Gao, Department of Social Medicine, Health Management College, Harbin 11 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected] 12 5 13 6 Mingli Jiao, Department of Health Policy, Health Management College, Harbin 14 15 7 Medical University, Harbin, Heilongjiang Province, China. Email: 16 17 8 [email protected] 18 For peer review only 19 9 Ning Ning, Department of Social Medicine, Health Management College, Harbin 20 21 10 Medical University, Harbin, Heilongjiang Province, China. Email: 22 23 11 [email protected] 24 25 12 Libo Liang, Department of Social Medicine, School of Public Health, Harbin Medical 26 27 13 University, Harbin, Heilongjiang Province, China. Email: [email protected] 28 29 14 Qunhong Wu, Department of Social Medicine, Health Management College, Harbin 30 31 15 Medical University, Harbin, Heilongjiang Province, China. Email: 32 33 16 [email protected] 34 35 17 Yanhua Hao, Department of Social Medicine, Health Management College, Harbin 36 18 Medical University, Harbin, Heilongjiang Province, China. Email: [email protected]

37 http://bmjopen.bmj.com/ 38 39 19 40 41 20 *Corresponding authors: 42 43 21 Qunhong Wu 44 22 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District,

45 on September 27, 2021 by guest. Protected copyright. 46 47 23 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 48 49 24 87502851; fax: + 86-0451-87502853 50 25 51 52 26 Yanhua Hao 53 54 27 Postal address: Harbin Medical University, 157 Baojian Road, Nangang District, 55 56 28 Harbin, Heilongjiang, 150086, China; E-mail: [email protected]; Tel.: + 86-0451- 57 58 29 87502860; fax: + 86-0451-87502853 59 60 30

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Keywords: Healthcare worker; Tertiary hospitals; Sleep problems; Influencing factors 5 6 2 Word counts 7 8 3 word count of the abstract: 296 9 4 word count of manuscript: 4092 10 11 5 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Sleep problems of healthcare workers in tertiary hospitals and 5 6 7 2 influencing factors identified through a multi-level analysis: a 8 9 3 cross-sectional study in China 10 11 12 4 13 14 15 5 Abstract 16 17 6 Objectives: Sleep problems are common among healthcare workers, especially among 18 For peer review only 19 20 7 those who work in tertiary hospitals; however, studies focusing on healthcare workers 21 22 8 in tertiary hospitals in China are limited. Therefore, we aimed to examine the 23 24 25 9 prevalence of sleep problems among tertiary hospital employees in China and identify 26 27 10 associated factors. 28 29 30 11 Design: Cross-sectional questionnaire survey study. 31 32 33 12 Setting: The study was conducted in 317 departments at 33 tertiary hospitals in the 34 35 13 Heilongjiang Province, China. 36

37 http://bmjopen.bmj.com/ 38 14 Participants: The study included a representative sample of 4,007 healthcare workers, 39 40 15 including clinicians, medical technicians, and office workers, who were selected using 41 42 43 16 a multi-stage cluster sampling method. Ultimately, 3,810 participants completed valid 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 questionnaires, yielding a response rate of 95%. 47 48 18 Primary and secondary outcome measures: Participants completed a structured 49 50 51 19 questionnaire, which collected data regarding demographics, sleep problems, 52 53 20 employment, chronic pain, and health-related behaviors. A two-level logistic regression 54 55 56 21 model was constructed to examine determinants of sleep problems. 57 58 59 22 Results: We found that nearly half of the respondents reported experiencing sleep 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 problems. The two-level logistic model suggested a positive association of sleep 5 6 7 2 problems with headaches (OR = 2.64, 95% CI = 2.27–3.07), working hours (OR = 1.41, 8 9 3 95% CI = 1.20–1.66), and night shifts (OR = 1.43, 95% CI = 1.21–1.69). In contrast, 10 11 12 4 regular diet (OR = 0.44, 95% CI = 0.36–0.53) and exercise (OR = 0.65, 95% CI = 0.53– 13 14 5 0.80) protected the study population against sleep problems. 15 16 17 6 Conclusions: We found that sleep problems are prominent among tertiary hospital 18 For peer review only 19 20 7 employees in China. The risk factor most strongly associated with sleep problems was 21 22 8 headaches. To alleviate sleep problems in this population, medical institutions and 23 24 25 9 health-policy makers should establish reasonable working schedules and night-shift 26 27 10 work systems, while the healthcare workers themselves should pay attention to their 28 29 30 11 chronic pain problems and engage in regular diet and exercise. 31 32 33 12 34 35 13 Strengths and limitations of this study 36

37 http://bmjopen.bmj.com/ 38 14  This was a multi-center study with a large sample size; specifically, we recruited 39 40 15 3,810 healthcare workers from 317 departments at 33 tertiary hospitals in the 41 42 43 16 Heilongjiang Province, China, using multi-stage stratified random sampling. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17  Considering the hierarchical structure of the study sample, a two-level logistic 47 48 18 model was applied to examine determinants of sleep problems in order to obtain 49 50 51 19 more accurate and reliable results. 52 53 20  Although the results of the study may not be generalizable to healthcare workers in 54 55 56 21 other countries, the findings likely accurately reflect the sleep problems of many 57 58 59 22 Chinese healthcare workers, particularly those working in tertiary public hospitals 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 in China. 5 6 7 2  Information regarding the exposure and outcome was obtained via a self-report 8 9 3 questionnaire, which may have introduced report bias. 10 11 12 4  The cross-sectional design limited our ability to establish causal relationships based 13 14 5 on the results. 15 16 17 6 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Introduction 5 6 7 2 Sleep problems are increasingly becoming a public health concern. Globally, it is 8 9 3 estimated that 7.6% of the population has sleep problems 1. Common sleep problems 10 11 12 4 include difficulty falling asleep, frequent nighttime awakenings, parasomnias, noisy 13 14 5 breathing, snoring, and obstructive sleep apnea 2. In China, workers in a variety of 15 16 17 6 occupations are at risk of sleep problems, although sleep problems are particularly 18 For peer review only 19 3 20 7 evident among healthcare workers . Many scholars have reported a high prevalence of 21 22 8 sleep problems among Chinese healthcare workers, ranging from 21% to 65.5% 4-6. In 23 24 25 9 fact, it has been shown that sleep problems are common among healthcare workers 26 27 10 worldwide, including clinicians, nurses, medical technicians, and administrative staff. 28 29 30 11 For example, in a study by Hayashino et al. 30.7% of Japanese healthcare workers 31 32 7 33 12 reported having difficulty initiating sleep . Moreover, a study performed in Italy found 34 35 13 that the prevalence of sleep disturbances among healthcare workers was extremely high, 36

37 http://bmjopen.bmj.com/ 38 14 even among healthy controls in the study, with 24.6% of workers reporting difficulties 39 40 15 falling asleep and 31.7% experiencing frequent awakenings during the night 8. In a 41 42 43 16 study in Iran, nearly 43% of healthcare workers reported different degrees of sleep 44

45 9 on September 27, 2021 by guest. Protected copyright. 46 17 problems . 47 48 18 Paying special attention to sleep problems among healthcare workers is essential for 49 50 51 19 several reasons. First, sleep problems can affect the health and safety of these workers. 52 53 20 For instance, various studies have confirmed that sleep problems have a powerful 54 55 56 21 influence on the risk of cardiovascular disease, immune disorders, and cancer 10-12. Sun 57 58 59 22 and colleagues reported a significant negative correlation between anxiety and sleep 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 problems in healthcare workers 4. Moreover, Kling et al. reported that sleep problems 5 6 13 7 2 are associated with an increased risk of work injury . Second, sleep problems affect 8 9 3 both the health of the healthcare workers and patient safety and the quality of medical 10 11 12 4 services the healthcare workers provide. Indeed, there is clear evidence indicating that 13 14 5 sleep deprivation caused by long working hours can affect both the health of healthcare 15 16 17 6 workers and the safety of patients 14. Further, sleep deprivation, long working hours, 18 For peer review only 19 20 7 and working during the night have been found to lead to medical errors, such as 21 22 8 increases in the rates of surgical complications, needlestick injuries, adverse drug 23 24 25 9 events, and misdiagnosis 15-18. 26 27 10 Currently, there is tremendous pressure on the healthcare system in China given the 28 29 30 11 increasing size of the aging population 19 and the elevated incidence of chronic diseases 31 32 20 21 33 12 such as cancer and cardiovascular diseases . In response to the increasing demands 34 35 13 on healthcare services, the workloads and hours worked by healthcare workers have 36

37 http://bmjopen.bmj.com/ 38 14 likewise increased, which may also lead to more frequent sleep problems in these 39 40 15 individuals. One prior study with healthcare workers indicated that long working hours 41 42 43 16 were the critical determinant of sleep health 22. However, few studies have focused on 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 the relationship between working hours and sleep problems in healthcare workers in 47 48 18 China. Moreover, Kim et al. suggested that frequent night shifts are likely to be an 49 50 51 19 important factor affecting sleep problems among healthcare workers 23. 52 53 20 Furthermore, chronic pain, which is common among healthcare workers 24, has been 54 55 56 21 shown to be related to sleep problems 25, although it remains unclear which specific site 57 58 59 22 of chronic pain most greatly affects sleep problems. Health-related behaviors are also 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 an important potential influencing factor. For example, research has shown that 5 6 7 2 unhealthy diets are associated with sleep problems, while a healthy diet can effectively 8 9 3 improve sleep quality 26. A recent study conducted in Japan observed that healthcare 10 11 12 4 workers who reported having trouble initiating sleep consumed more alcohol and 13 14 5 exercised less 7. In addition, it has been noted that sleep health assessment should be 15 16 17 6 multidimensional, with sleep duration being one of the most important aspects 27 28. 18 For peer review only 19 20 7 Although sleep needs differ among individuals due to genetic and environmental factors 21 22 8 29, shorter sleep duration may increase the likelihood of sleep problems 30. 23 24 25 9 In 2009, China launched a nationwide systematic reform of healthcare services. 26 27 10 Although the reform was successful in some areas, additional steps are required to 28 29 30 11 reform the delivery of China’s healthcare services. Particularly, primary healthcare 31 32 33 12 facilities have not been able to perform a gate-keeping function, and most patients in 34 35 13 China prefer to visit tertiary hospitals, especially tertiary public hospitals, even for 36

37 http://bmjopen.bmj.com/ 38 14 common and minor illnesses. In 2016, there were approximately 3.27 billion patient 39 40 15 visits to Chinese hospitals, 49.8% of which were visits to tertiary hospitals 31. A survey 41 42 43 16 of the Chinese Medical Doctor Association in 2014 showed that 92% of healthcare 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 workers in tertiary hospitals had to work overtime, and 72% of the healthcare workers 47 48 18 who work > 60 h a week on average are employed in tertiary hospitals in China 32. 49 50 51 19 Therefore, the sleep problems of healthcare workers in tertiary hospitals are of 52 53 20 paramount importance. Nevertheless, population-based prevalence data on sleep 54 55 56 21 problems among healthcare workers in tertiary hospitals in China are limited. Further, 57 58 59 22 the relationships among sleep problems, work factors, health-related behaviors, and 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 chronic pain remain to be discovered in this population. 5 6 7 2 Therefore, the purposes of the present study were to investigate the prevalence of self- 8 9 3 reported sleep problems among healthcare workers in tertiary public hospitals in China 10 11 12 4 and to identify potential factors associated with sleep problems among these individuals. 13 14 5 This study has several advantages. First, although it has been established that sleep 15 16 17 6 problems are common among healthcare workers, relevant studies examining tertiary 18 For peer review only 19 20 7 hospital employees in China are limited. Second, considering the hierarchical structure 21 22 8 of the study sample, a two-level logistic model was applied to examine determinants of 23 24 25 9 sleep problems in order to obtain more accurate and reliable results. Third, in contrast 26 27 10 to previous reports 33-35, this was a multi-center study covering a large sample of 28 29 30 11 healthcare workers in the Heilongjiang Province of China. 31 32 12 33 34 13 Methods 35 36 14 Data collection and study population

37 http://bmjopen.bmj.com/ 38 39 15 This study took place in Heilongjiang Province, in Northeast China, where there were 40 41 36 42 16 689 public hospitals in 2014, of which 82 were tertiary public hospitals . The study 43 44 17 included a representative sample of healthcare workers from tertiary hospitals,

45 on September 27, 2021 by guest. Protected copyright. 46 47 18 including clinicians (physicians, surgeons, pediatricians, gynecologists, emergency 48 49 50 19 doctors, etc.), medical technicians (pharmacists, radiological technicians, laboratory 51 52 20 physicians, etc.), and office workers (administrative department personnel, finance 53 54 55 21 personnel, etc.), who were selected using a multi-stage cluster sampling method. First, 56 57 22 33 tertiary public hospitals were selected at random to serve as survey locations for this 58 59 60 23 study and all selected hospitals participated (100% response rate). Second, the

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 departments of each hospital were coded, and several departments were randomly 5 6 7 2 selected from each hospital. The sample size at each level was based on the proportion 8 9 3 of the number of workers in the hospital at that level. In the final sample, 4,007 10 11 12 4 healthcare workers from 317 departments were invited to participate in the study. The 13 14 5 healthcare workers in the selected departments completed a self-administered 15 16 17 6 questionnaire (see Additional file 1 for details). After the data cleaning process, 18 For peer review only 19 20 7 incomplete or anomalous data (n = 197) were excluded, leaving a final sample of 3,810 21 22 8 participants (95% response rate). 23 24 25 9 26 27 28 10 Variables 29 30 11 A structured questionnaire was designed to obtain data regarding demographics, sleep 31 32 33 12 problems, working conditions, chronic pain, and health-related behaviors. 34 35 13 36

37 http://bmjopen.bmj.com/ 38 14 Outcome measure (dependent variable) 39 40 15 The main outcome was the presence of sleep problems, which was a binary variable (1 41 42 43 16 = yes; 0 = no). The presence of sleep problems was assessed from each participant’s 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 self-report. Participants were asked the question “Overall, in the last 30 days, how much 47 48 18 of a problem did you have with sleeping, such as falling asleep, waking up frequently 49 50 51 19 during the night, or waking up too early in the morning?” with the following five 52 53 20 response options: none, mild, moderate, severe, and extreme. Participants who 54 55 56 21 answered “extreme” and “severe” were considered to have sleep problems. This 57 58 59 22 definition was adopted by previous studies that used the same survey question to assess 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 sleep problems 1 37. 5 6 7 2 8 9 3 Individual-level variables (independent variables) 10 11 12 4 Individual-level variables included sociodemographic indicators (age, sex, educational 13 14 5 background, job title, and monthly salary), working conditions (average weekly 15 16 17 6 working hours, number of night shifts per month, and one-way commute time), health- 18 For peer review only 19 20 7 related behaviors (physical exercise and regularity of diet), and experience of chronic 21 22 8 pain (headache, neck and shoulder pain, lumbar and back pain, and knee pain). 23 24 25 9 Age was categorized into three groups (< 30, 30–40, and > 40 years) and job title was 26 27 10 considered an ordinal categorical variable (junior title or no title, intermediate job title, 28 29 30 11 and senior title). The median monthly salary of participants in this study was 3,000 31 32 33 12 yuan. Therefore, we coded monthly salary as ≤ 3,000 yuan (=0) and > 3,000 yuan (=1). 34 35 13 Average weekly working hours were evaluated with the question: “How much time do 36

37 http://bmjopen.bmj.com/ 38 14 you work on average per week?” The median number of weekly working h reported by 39 40 15 participants in this study was 55. Hence, the response categories were 1 = ≥ 55 h and 0 41 42 43 16 = < 55 h. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 Herein, chronic pain was defined as any continuous or persistent intermittent pain 47 48 18 experienced for a period longer than 3 months 38. We asked participants whether they 49 50 51 19 had experienced chronic pain that had not improved in the past 30 days in the following 52 53 20 regions: headache, neck and shoulder, lumbar and back, and knee. The response 54 55 56 21 categories were 1 = yes and 0 = no. The coding of other individual-level variables is 57 58 59 22 shown in Additional file 2. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 5 6 7 2 Department-level variables (independent variable) 8 9 3 The participants’ department type was a level-2 explanatory variable in our analysis 10 11 12 4 because individuals in the same department are not independent of each other. Among 13 14 5 the healthcare workers in the study, some practitioners were from clinical departments 15 16 17 6 (including internal medicine, surgery, gynecology, pediatrics, etc.) or laboratory, 18 For peer review only 19 20 7 pathology, radiology, or other, technicians. Moreover, administrative-department 21 22 8 personnel also participated in the survey. Therefore, this study was based on the 23 24 25 9 functional attributes of the department, and whether the participants were from a 26 27 10 clinical department (1 = yes, 0 = no) was an explanatory variable. 28 29 30 11 31 32 33 12 Data analysis 34 35 13 In this study, we used two statistical methods. The chi-squared test was used to examine 36

37 http://bmjopen.bmj.com/ 38 14 the independent associations between participant sleep problems and each independent 39 40 15 variable. As the data were hierarchical, participants (level-1) were nested within 41 42 43 16 departments (level-2), to create a two-level logistic model that was applied to assess 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 associations between sleep problems and the study variables. 47 48 18 Intra-class correlation coefficients (ICC) for healthcare workers were calculated to 49 50 51 19 assess clustering by department 39. First, an empty model was constructed to verify 52 53 20 whether our research model was applicable to the two-level logistic model. A 54 55 56 21 significant ICC would suggest that a multi-level logistic regression model is appropriate. 57 58 59 22 Subsequently, in model 1, the department-level explanatory variable was added to the 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 empty model. Next, in model 2, the individual-level explanatory variables were added 5 6 7 2 to model 1. Meanwhile, we needed to judge whether the two multi-category variables 8 9 3 (age and job title) should be introduced into the model as dummy or grouped linear 10 11 12 4 variables according to the fit of the model. The likelihood ratio (LR) chi-square test can 13 14 5 be used for model comparisons and the difference in -2 log-likelihood between two 15 16 17 6 models is used for the LR chi-square test 40. A statistically significant difference in -2 18 For peer review only 19 20 7 log-likelihood between the two models indicated that the fit of the latter model was 21 22 8 better 41. We also conducted subpopulation analyses to determine whether short sleep 23 24 25 9 duration (Q15: 1 = sleep duration ≤ 6 h, 0 = sleep duration > 6 h) 42 significantly 26 27 10 modified the observed associations 43. 28 29 30 11 The two-level binary logistic regression models were estimated with the GLIMMIXED 31 32 33 12 and NLMIXED procedures of SAS 9.4 (SAS Institute Inc., Cary, NC, USA). All other 34 35 13 analyses were conducted with SPSS 21.0 (IBM Corp., Armonk, NY, USA), and p 36

37 http://bmjopen.bmj.com/ 38 14 values < 0.05 were considered to reflect statistical significance. 39 40 15 41 42 43 16 Ethics approval 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 The study protocol was reviewed and approved by the Research Ethics Committee of 47 48 18 Harbin Medical University (project identification code: HMUIRB20170016) and the 49 50 51 19 ethics committee from each hospital (see the Additional file 3). The study also 52 53 20 received permission and assistance from each hospital to contact their healthcare 54 55 56 21 workers. All participants were informed that the study was voluntary and confidential 57 58 59 22 and that they had the right to discontinue their participation or withdraw their consent 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 at any time without obligation to state their reasons. Interviews were conducted after 5 6 7 2 the participants provided written informed consent. Survey data were collected using 8 9 3 an anonymous questionnaire to protect the privacy of the participants. 10 11 12 4 13 14 15 5 Patient and public involvement 16 17 6 No patients were involved. 18 For peer review only 19 20 7 21 22 8 Results 23 24 25 9 Participant characteristics 26 27 10 The participants’ general characteristics are shown in Table 1. Overall, participants 28 29 30 11 were generally unable to maintain a regular diet (75.5%). The most common types of 31 32 33 12 chronic pain were neck and shoulder pain (69.9%), and lumbar and back pain (64.1%). 34 35 13 Sleep problems were reported by 48.5% of participants. The univariate analysis 36

37 http://bmjopen.bmj.com/ 38 14 indicated that demographics (excluding sex and monthly salary), chronic pain, and 39 40 15 working conditions were all related to participant sleep problems (p < 0.05). 41 42 43 16 44

45 17 Table 1. Characteristics of questionnaire respondents and results of the univariate on September 27, 2021 by guest. Protected copyright. 46 18 analysis 47 48 Variable Total Sleep problems p-value 49 50 N = 3810 (%) No Yes 51 n =1964 (%) n = 1846 (%) 52 Department-level 53 54 Clinical department .001 55 Yes 3,297 (86.5) 1,658 (50.2) 1,639 (49.7) 56 No 513 (13.5) 306 (59.6) 207 (40.4) 57 Individual-level 58 59 Age (years) < .001 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 42

1 2 3 < 30 504 (13.2) 308 (61.1) 196 (38.9) BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 30–40 2,054 (53.9) 982 (47.8) 1,072 (52.2) 6 > 40 1,252 (32.9) 674 (53.8) 578 (46.2) 7 Sex .279 8 9 Men 2,455 (64.4) 1,282 (52.2) 1,173 (47.8) 10 Women 1,355 (35.6) 682 (50.3) 673 (49.7) 11 12 Educational background .004 13 Bachelor’s degree and below 1,883 (49.4) 926 (49.2) 957 (50.8) 14 15 Master's degree and above 1,927 (50.6) 1,038 (53.9) 889 (46.1) 16 Job title .001 17 18 Junior titles and othersFor peer924 review (24.3) 521 only (48.8) 403 (51.2) 19 Intermediate title 1,287 (33.8) 622 (30.8) 665 (69.2) 20 Senior title 1,599 (42.0) 821 (51.3) 778 (48.7) 21 Monthly salary .791 22 23 ≤ 3,000 yuan 2,286 (60.0) 1,112 (48.6) 1,174 (51.4) 24 > 3,000 yuan 1,524 (40.0) 790 (51.8) 734 (48.2) 25 One-way commute time .001 26 27 < 30 minutes 2,038(53.5) 1097 (53.8) 941 (46.2) 28 ≥ 30 minutes 1,772(46.5) 867 (48.9) 905 (51.1) 29 Regular diet at work < .001 30 31 Yes 934 (24.5) 682 (73.0) 252 (27.0) 32 No 2,876 (75.5) 1,282 (44.6) 1,594 (55.4) 33 34 No 3,063 (80.4) 1,427 (34.2) 1,636 (65.8) 35 Work hours per week (hours) < .001 36 ≥ 55 1,964 (51.5) 850 (42.3) 1,114 (56.7)

37 http://bmjopen.bmj.com/ 38 < 55 1,846 (48.5) 1,114 (60.3) 732 (39.7) 39 Night shifts per month (times) < .001 40 41 ≤ 5 2,171 (57.0) 1,270 (58.5) 901 (41.5) 42 > 5 1,639 (43.0) 694 (42.3) 945 (57.7) 43 44 Daily exercise < .001

45 Yes 685 (18.0) 452 (66.0) 233 (34.0) on September 27, 2021 by guest. Protected copyright. 46 No 3,125 (82.0) 1,512 (48.4) 1,613 (51.6) 47 48 Headaches < .001 49 Yes 1,698 (44.6) 601 (35.4) 1,097 (64.6) 50 51 No 2,112 (55.4) 1,363 (64.5) 749 (35.5) 52 Neck and shoulder pain < .001 53 54 Yes 2,664 (69.9) 1,243 (46.7) 1,421 (53.3) 55 No 1,146 (30.1) 721 (62.9) 425 (37.1) 56 Lumbar and back pain <.001 57 58 Yes 2,441 (64.1) 1,111 (45.5) 1,330 (54.5) 59 No 1,369 (35.9) 853 (62.3) 516 (37.7) 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 42 BMJ Open

1 2 3 Knee pain <.001 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Yes 975 (25.6) 340 (34.9) 635 (65.1) 6 No 2,835 (74.4) 1,624 (57.3) 1,211 (42.7) 7 8 1 9 10 2 Self-reported sleep duration 11 12 3 Most participants (46.7%) reported a sleep duration of approximately 6 h per day 13 14 15 4 regardless of department. Moreover, most of the 11.3% of respondents who reported 16 17 18 5 sleeping less thanFor 5 h per peer day were working review in clinical onlydepartments (Figure 1). 19 20 6 21 22 23 7 Multi-level analysis 24 25 8 Table 2 presents the results from the multi-level analyses. The results of the 26 27 28 9 empty model showed significant between-group variation (ICC = 0.1181, p < 0.001). 29 30 31 10 An ICC of 0.1181 indicated that approximately 12% of the total variation in the 32 33 11 outcome variables was caused by variation between the departments. 34 35 36 12 Subsequently, the department-level explanatory variable was introduced. The results

37 http://bmjopen.bmj.com/ 38 13 showed that the odds of having sleep problems was approximately 1.5 times greater 39 40 41 14 among healthcare workers in clinical departments than among those in non-clinical 42 43 44 15 departments (OR = 1.53, 95% CI = 1.15–2.04). After incorporating the department-

45 on September 27, 2021 by guest. Protected copyright. 46 16 level explanatory variable, the -2 log-likelihood declined from 5,159.4 to 5,150.9. The 47 48 49 17 difference in -2 log-likelihood between model 1 and the empty model indicated that 50 51 18 the fit of model 1 was better (LR = 8.5, df = 1, p < 0.01). 52 53 54 19 In model 2, individual-level variables were added. The results showed that healthcare- 55 56 57 20 worker sleep problems had an inverse association with regular diet (OR = 0.44, 95% 58 59 21 CI = 0.36–0.53) and exercise (OR = 0.65, 95% CI = 0.53–0.80). The odds of having 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 sleep problems were higher among participants with senior than with junior job titles 5 6 7 2 (OR = 0.71, 95% CI = 0.55–0.92). Model 2 further suggested that chronic pain was 8 9 3 positively associated with sleep problems. Specifically, participants with sleep 10 11 12 4 problems were more likely to report having headaches (OR = 2.64, 95% CI = 2.27– 13 14 5 3.03). Furthermore, working hours (OR = 1.41, 95% CI = 1.20–1.66) and night shifts 15 16 17 6 (OR = 1.43, 95% CI = 1.21–1.69) were associated with sleep problems. The 18 For peer review only 19 20 7 difference in -2 log-likelihood between model 2 and model 1 indicated that the fit of 21 22 8 model 2 was better (LR = 612, df = 14, p < 0.001). Meanwhile, department-level 23 24 25 9 variables were also tested but showed no statistical significance. 26 27 10 28 11 Table 2. Factors related to sleep problems: results from the two-level logistic regression 29 12 analysis (N = 3,810) 30 31 Variable Empty model Model 1 Model 2 32 Department-level 33 Clinical department 34 ** 35 Yes 1.53 (1.15, 2.04) 1.07 (0.81, 1.40) 36 No 1 (reference) 1 (reference)

37 Individual-level http://bmjopen.bmj.com/ 38 39 Age (years) 40 < 30 0.87 (0.63, 1.20) 41 30–40 1.06 (0.86, 1.31) 42 43 > 40 1 (reference) 44 Educational background

45 Bachelor degree and below 0.87 (0.74, 1.03) on September 27, 2021 by guest. Protected copyright. 46 47 Master's degree and above 1 (reference) 48 Job title 49 Junior title or no title 0.71 (0.55, 0.92) ** 50 Intermediate title 0.87 (0.71, 1.07) 51 52 Senior title 1 (reference) 53 One-way commute time 54 <30 minutes 1.07 (0.92, 1.24) 55 56 ≥30 minutes 1 (reference) 57 Regular diet at work 58 Yes 0.44 (0.36, 0.53) *** 59 60 No 1 (reference)

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 42 BMJ Open

1 2 3 Daily exercise BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 *** 5 Yes 0.65 (0.53, 0.80) 6 No 1 (reference) 7 Work hours per week 8 *** 9 ≥ 55 1.41 (1.20, 1.66) 10 < 55 1 (reference) 11 Night shifts per month 12 *** 13 ≤ 5 1.43 (1.21, 1.69) 14 > 5 1 (reference) 15 Headache 16 Yes 2.64 (2.27, 3.07) *** 17 18 No For peer review only 1 (reference) 19 Neck and shoulder pain 20 Yes 1.29 (1.09, 1.52) ** 21 22 No 1 (reference) 23 Lumbar and back pain 24 *** 25 Yes 1.36 (1.16, 1.60) 26 No 1 (reference) 27 Knee pain 28 Yes 1.61 (1.34, 1.93) *** 29 30 No 1 (reference) 31 Sigma μ (SE) 0.4407 (0.092) 0.4061 (0.088) 0.2278 (0.070) 32 Intra-class correlation 0.1181*** 0.1099*** 0.0648*** 33 34 −2 log-likelihood 5159.4 5150.9 4538.0 35 1 * p < 0.05; ** p < 0.01; *** p < 0.001 36

37 2 http://bmjopen.bmj.com/ 38 39 40 3 Table 3 shows the results of the two-level logistic regression model for the 41 42 4 subpopulation of healthcare workers with short sleep durations. In this subpopulation, 43 44 5 the observed associations were mostly consistent with those identified with the two-

45 on September 27, 2021 by guest. Protected copyright. 46 47 6 level model comprising the total sample. However, in contrast to the analysis of the 48 49 50 7 total sample, neck and shoulder pain (p = 0.128) was not significantly associated with 51 52 53 8 sleep problems in healthcare workers with short sleep durations. 54 55 9 56 10 Table 3. Factors related to sleep problems: results from the two-level logistic regression 57 58 11 analysis of healthcare workers with short sleep durations (≤6 h; n = 2,210) 59 Variable Empty model Model 1 Model 2 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 42

1 2 3 Department-level BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 Clinical department 6 Yes 1.47(1.21, 1.79) * 0.86 (0.59, 1.25) 7 No 1 (reference) 1 (reference) 8 9 Individual-level 10 Age (years) 11 < 30 0.92 (0.60, 1.42) 12 13 30–40 1.11 (0.84, 1.46) 14 > 40 1 (reference) 15 Educational background 16 17 Bachelor degree and below 0.86 (0.69, 1.06) 18 Master's degree andFor above peer review only 1 (reference) 19 Job title 20 ** 21 Junior title or no title 0.63 (0.45, 0.88) 22 Intermediate title 0.84 (0.63, 1.11) 23 Senior title 1 (reference) 24 One-way commute time 25 26 <30 minutes 1.13 (0.93, 1.38) 27 ≥30 minutes 1 (reference) 28 Regular diet at work 29 *** 30 Yes 0.53 (0.40, 0.71) 31 No 1 (reference) 32 Daily exercise 33 * 34 Yes 0.74 (0.56, 0.98) 35 No 1 (reference) 36 Work hours per week

37 ≥ 55 1.32 (1.06, 1.63) *** http://bmjopen.bmj.com/ 38 39 < 55 1 (reference) 40 Night shifts per month 41 ≤ 5 1.45 (1.17, 1.81) ** 42 43 > 5 1 (reference) 44 Headache *** 45 Yes 2.58 (2.11, 3.16) on September 27, 2021 by guest. Protected copyright. 46 47 No 1 (reference) 48 Neck and shoulder pain 49 Yes 1.19 (0.95, 1.49) 50 No 1 (reference) 51 52 Lumbar and back pain 53 Yes 1.29 (1.04, 1.60) * 54 55 No 1 (reference) 56 Knee pain 57 Yes 1.63 (1.29, 2.05) *** 58 59 No 1 (reference) 60 Sigma μ (SE) 0.3885 (0.100) 0.3842 (0.099) 0.2789 (0.098)

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 42 BMJ Open

1 2 3 Intra-class correlation 0.1056*** 0.1046*** 0.07814*** BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 −2 log-likelihood 2922.1 2920.5 2645.1 6 1 * p < 0.05; ** p < 0.01; *** p < 0.001 7 8 2 9 10 11 3 Factors related to headaches 12 13 4 Table 4 shows factors related to headaches among respondents with sleep problems. 14 15 16 5 We found that among the participants with headache, 62.4% worked ≥ 55 h per week 17 18 6 and 73.5% did notFor exercise peer daily. review only 19 20 21 7 22 23 8 Table 4. Factors related to headaches among respondents with sleep problems (n = 24 9 1,846) 25 26 Variable Headache p-value 27 Yes No 28 n= 1097 (%) n= 749 (%) 29 30 Work hours per week .015 31 ≥55 hours 685(62.4) 429(57.3) 32 33 <55 hours 412(37.6) 320(42.7) 34 Number of night shifts per .173 35 month 36 5 times or less 525(47.9) 376(50.2)

37 http://bmjopen.bmj.com/ 38 More than 5 times 572(52.1) 373(49.8) 39 40 Do exercise every day < .001 41 Yes 120 (10.9) 113(15.1) 42 No 977 (89.1) 636(84.9) 43 44 Regular diet at work < .001

45 Yes 122 (11.12) 130 (17.36) on September 27, 2021 by guest. Protected copyright. 46 47 No 975 (88.88) 619 (82.64) 48 Neck and shoulder pain < .001 49 50 Yes 880 (80.2) 540 (72.1) 51 No 217 (19.8) 209 (27.9) 52 Lumbar and back pain < .001 53 54 Yes 833 (75.9) 496 (66.2) 55 No 264 (24.1) 253 (33.8) 56 57 Knee pain < .001 58 Yes 498 (45.4) 136 (18.2) 59 No 559 (54.6) 613 (81.8) 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 5 6 7 2 Discussion 8 9 3 This study contributes to current knowledge on the extent of sleep problems among 10 11 12 4 healthcare workers. We determined the prevalence of sleep problems experienced by 13 14 5 the healthcare employees of tertiary public hospitals in the Heilongjiang Province, 15 16 17 6 China, and identified associated factors. In this study, more than 45% of the sample 18 For peer review only 19 20 7 reported experiencing sleep problems, which is much higher than the prevalence 21 22 8 reported for the general population of China (36.2%) 3. 23 24 25 9 Herein, the factor most strongly associated with sleep problems among healthcare 26 27 10 workers was headaches. We found that 44.6% of participants reported experiencing 28 29 30 11 headaches for longer than 3 months, which had not improved in the past 30 days. 31 32 33 12 Such findings are consistent with those of previous studies showing that headache is a 34 35 13 very common symptom among healthcare workers. Sokolovic et al. found that 61% of 36

37 http://bmjopen.bmj.com/ 38 14 the healthcare workers they evaluated had experienced at least one type of headache 39 40 15 in the previous 3 months. A cross-sectional study with healthcare workers conducted 41 42 43 16 in Enugu, South East Nigeria, showed that 12% of the participants had severe 44

45 44 on September 27, 2021 by guest. Protected copyright. 46 17 headaches that affected their sleep . From a clinical perspective, it has been 47 48 18 suggested that the relationship between headaches and sleep problems may be 49 50 51 19 bidirectional 45, whereby headache treatment is crucial for relieving sleep problems, 52 53 20 while sleep improvement is also likely to reduce headache symptoms in healthcare 54 55 56 21 workers. In addition, we analyzed factors related to the participants’ headaches and 57 58 59 22 found that among respondents with sleep problems, irregular diet, working overtime, 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 lack of exercise, and neck and shoulder, lumbar and back, and knee pain were 5 6 7 2 associated with headaches. These findings highlight the importance of addressing 8 9 3 these factors to reduce the high incidence of headaches among healthcare workers. 10 11 12 4 Our study additionally confirmed the negative effects of other types of chronic pain 13 14 5 on sleep, suggesting that chronic pain in different parts of the body increases the 15 16 17 6 likelihood of sleep problems. A study by Andersen et al. with healthcare workers also 18 For peer review only 19 20 7 indicated that chronic musculoskeletal pain in one body region increases the risk of 21 22 8 developing chronic pain in other pain-free body regions 24. Therefore, when 23 24 25 9 healthcare workers complain of chronic pain, they should be diagnosed and treated in 26 27 10 a timely manner to reduce the risk of sleep problems and avoid the development of 28 29 30 11 chronic pain in other body parts. 31 32 33 12 Our finding that night-shift work predicted a greater risk of sleep problems is 34 35 13 consistent with the results of several previous studies 9 34. In hospitals, inpatient care 36

37 http://bmjopen.bmj.com/ 38 14 usually requires round-the-clock services, which often translates to healthcare workers 39 40 15 having to work extended shifts (especially night shifts) and long hours, which deprive 41 42 43 16 them of sleep. Moreover, working during the night leads to misalignment between 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 internal circadian rhythms and work–rest schedules, again resulting in fewer sleeping 47 48 18 hours 46. As such, determining effective means to minimize the negative effects of 49 50 51 19 shift work on health workers’ sleep is essential. Previous studies have suggested that 52 53 20 timed bright lights and light-blocking goggles can promote circadian adaptation to 54 55 56 21 shift work 47 48. However, more reasonable night-shift working schedules may be a 57 58 59 22 better solution to alleviate the sleep problems of healthcare employees; our data 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 suggest that the number of night shifts per month should be reduced at a reasonable 5 6 7 2 level. Additionally, and in line with the assertions of some previous studies, we 8 9 3 suggest that it would be optimal to allow for time off of sufficient duration between 10 11 12 4 shifts 49 and to set a nap time to alleviate the high levels of sleepiness experienced 13 14 5 while working during the night 49 50. 15 16 17 6 The present study also revealed that long working hours were a critical risk factor 18 For peer review only 19 20 7 associated with sleep problems. A cross-sectional study in the United States reported 21 22 8 that compensated working hours were consistently associated with sleep duration, 23 24 25 9 such that as work duration increased, sleep duration decreased 51. In addition, the 26 27 10 working hours of healthcare workers are evidently too long. According to the latest 28 29 30 11 survey data collected in 2017, tertiary hospital employees in China work an average 31 32 52 33 12 of 51.05 h per week , which is similar to the hours reported by our participants. Hale 34 35 13 et al. demonstrated that people working > 50 h per week were more likely to be short 36

37 http://bmjopen.bmj.com/ 38 14 sleepers 53. Many countries have clear rules limiting the maximum working hours of 39 40 15 healthcare workers. For example, in 2009, the European Union specified a maximum 41 42 43 16 of 48 h of work per week and established rest periods for healthcare workers 54. 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 Although China’s labor law specifies a maximum number of working hours, it does 47 48 18 not apply to healthcare workers, and currently, there are no established policies 49 50 51 19 concerning the labor rights of healthcare workers 55. 52 53 20 Our findings confirmed that health-related behaviors play an important role in sleep 54 55 56 21 problems. Indeed, our results showed that more than three-quarters of healthcare 57 58 59 22 workers were unable to maintain a regular diet, and an irregular diet was significantly 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 associated with sleep problems. Many healthcare workers do not have enough time to 5 6 7 2 eat while working, and thus they order fast food from delivery services or simply 8 9 3 delay their meals. However, research has revealed that short sleep duration is 10 11 12 4 associated with higher caloric intake, mostly due to increased consumption of 13 14 5 saturated fat 56. We recommend that healthcare workers should maintain a regular diet 15 16 17 6 and increase the intake of micronutrients such as tryptophan and vitamins B to 18 For peer review only 19 57 20 7 improve sleep problems . In addition, we also observed that participants who did not 21 22 8 engage in regular exercise were more likely to experience sleep problems, which is 23 24 25 9 consistent with the findings of previous research 58. A randomized controlled trial 26 27 10 found that improved sleep quality was evident after an exercise intervention and that 28 29 30 11 physical activity is beneficial in terms of quality and duration of sleep 57. For adults 31 32 33 12 aged 18–64 years, the World Health Organization recommends at least 150 min of 34 35 13 moderate-intensity or at least 75 min of vigorous-intensity aerobic physical activity 36

37 http://bmjopen.bmj.com/ 38 14 per week 59. 39 40 15 A study funded by the Chinese Medical Doctor Association found 46 cases of karoshi, 41 42 43 16 or “death by overwork,” among Chinese healthcare workers between 2013 and 2015, 44

45 60 on September 27, 2021 by guest. Protected copyright. 46 17 most of which were aged 30–39 years . Although healthcare workers in China’s top- 47 48 18 three hospitals treat numerous patients, few workers are mindful of their own health 49 50 51 19 issues. Studies have shown that continuous exposure to factors such as lack of sleep, 52 53 20 exercise, and a regular diet, as was found here, may increase the risk of chronic 54 55 56 21 diseases among healthcare workers 10 61 62. The fact that healthcare workers do not pay 57 58 59 22 proper attention to their own health should be recognized by the government and 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 society. 5 6 7 2 Overall, existing evidence indicates that it is important for healthcare workers to 8 9 3 increase their sleep duration and reduce their workload. The effect that chronic pain, 10 11 12 4 especially headache, has on healthcare worker sleep problems is of particular concern 13 14 5 and requires attention. Furthermore, ensuring that workers engage in healthy 15 16 17 6 behaviors is equally important. We make the following recommendations based on 18 For peer review only 19 20 7 our findings: (1) hospitals should ensure that healthcare worker chronic pain in 21 22 8 different body parts is diagnosed and treated in a timely manner; (2) healthcare 23 24 25 9 workers should be aware of their own health issues and pay more attention to 26 27 10 cultivating healthy behaviors, such as eating a regular diet and undergoing regular 28 29 30 11 physical examinations; (3) more manageable working hours and night-shift work 31 32 33 12 systems should be established; and (4) health administrative departments should 34 35 13 expand the scope of labor laws and formulate special regulations for the labor rights 36

37 http://bmjopen.bmj.com/ 38 14 of healthcare workers. 39 40 15 Our study has some limitations of note. First, information regarding the exposure and 41 42 43 16 outcome was obtained via a self-report questionnaire, which may have introduced 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 report bias in these variables. In particular, we determined sleep problems via 47 48 18 subjective evaluations of the participants. Future research should incorporate 49 50 51 19 validated scales and clinical diagnoses to support the results obtained via self-reports. 52 53 20 Second, all participants were from the Heilongjiang Province, China, and thus the 54 55 56 21 study findings may not be generalizable to other populations. Moreover, the sample 57 58 59 22 did not include nursing staff, and hence the results cannot be used to characterize all 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 healthcare workers. Third, as this was a cross-sectional study, we cannot establish 5 6 7 2 causal relationships based on the results. Fourth, only one subpopulation of healthcare 8 9 3 workers was examined to test for effect modification. Finally, survey weights were 10 11 12 4 not considered in this study, and the results were based on an unweighted analysis, the 13 14 5 odds ratio of which might be smaller than that when considering weight 63. Therefore, 15 16 17 6 sample weighting should be conducted in future analyses. 18 For peer review only 19 20 7 21 22 8 Conclusions 23 24 25 9 In summary, our findings suggest that nearly half of the healthcare workers in tertiary 26 27 10 hospitals from the Heilongjiang Province in China have sleep problems. Chronic pain 28 29 30 11 (headaches, knee pain, neck and shoulder pain, and lumbar and back pain), long 31 32 33 12 working hours, and frequent night shifts were positively associated with sleep 34 35 13 problems among healthcare workers. Of these, headaches were the most prominent 36

37 http://bmjopen.bmj.com/ 38 14 risk factor for sleep problems in this population. However, a regular diet and daily 39 40 15 exercise may reduce the likelihood of sleep problems. Most findings were not altered 41 42 43 16 when analyzing only participants with short sleep durations. These results indicate 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 that to improve the sleep problems of healthcare workers, both medical institutions 47 48 18 and health-policy makers should work hand in hand to provide healthcare workers 49 50 51 19 with reasonable work schedules, while the healthcare workers themselves should pay 52 53 20 attention to their chronic pain problems and engage in regular diet and exercise. 54 55 56 21 57 58 59 22 Acknowledgements 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 The authors are very grateful to Miaomiao Zhao, Ding Ding, Linghan Shan for 5 6 7 2 providing the suggestion to draft the manuscript. We also thank all the participating 8 9 3 organizations and participants. 10 11 12 4 13 14 15 5 Footnotes 16 17 6 Funding 18 For peer review only 19 20 7 This research was supported by the Think Tank of Public Health Security and Health 21 22 8 Reform of Heilongjiang Province and the CMB Distinguished Professorship Awards 23 24 25 9 (Grant No. G16916400). This work was also supported by the National Natural 26 27 10 Science Foundation of China (Grant No. 71333003). 28 29 30 11 31 32 33 12 Competing interests 34 35 13 The authors declare that they have no competing interests. 36

37 http://bmjopen.bmj.com/ 38 14 39 40 15 Contributors 41 42 43 16 HL conducted calculations, analyzed results and drafted the manuscript. QHW and 44

45 on September 27, 2021 by guest. Protected copyright. 46 17 YHH were responsible for the overall design of the research, organized and conducted 47 48 18 the survey. MLJ, NN and LL designed the analyses framework. ZK and LJG revised 49 50 51 19 the paper. JJL, MC, XT and TZ assisted with the literature review and data collection. 52 53 20 All authors approval of the current version of this manuscript for publications. 54 55 56 21 57 58 59 22 Patient consent 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 1 Patient consent was obtained for publication 5 6 7 2 8 9 3 Data sharing statement 10 11 12 4 No additional data are available. 13 14 5 15 16 17 6 References 18 For peer review only 19 7 1. Koyanagi A, Stickley A. The Association between Sleep Problems and Psychotic Symptoms in the 20 8 General Population: A Global Perspective. Sleep 2015;38(12):1875-85. doi: 21 22 9 10.5665/sleep.5232 [published Online First: 2015/06/19] 23 10 2. Owens JA, Mindell JA. A Clinical Guide to Pediatric Sleep. Journal of the American Academy of Child 24 11 & Adolescent Psychiatry 2005;44(7):720–21. 25 12 3. Chinese Medical Association. China Sleep Quality Index announced 2014. 2014. 26 27 13 4. Sun W, Fu J, Chang Y, et al. Epidemiological study on risk factors for anxiety disorder among 28 14 Chinese doctors. Journal of Occupational Health 2012;54(1):1-8. 29 15 5. He Tian, Yancheng Di, Song J. Investigation of sleep quality and influncing factors of medical 30 31 16 workers in a grade -A tertiary hospitals. Chinese Journal of Social Medicine 2015;32(1):48-51. 32 17 6. Yang X, Wang R, Liu Z, et al. Relationship between Sleep Quality and Mental Health of Physicians. 33 18 China Journal of Health Psychology 2007;15(7):654-56. 34 35 19 7. Hayashino Y, Fukuhara S, Suzukamo Y, et al. Relation between sleep quality and quantity, quality of 36 20 life, and risk of developing diabetes in healthy workers in Japan: the High-risk and Population

37 21 Strategy for Occupational Health Promotion (HIPOP-OHP) Study. BMC Public Health http://bmjopen.bmj.com/ 38 22 2007;7:129. doi: 10.1186/1471-2458-7-129 [published Online First: 2007/06/29] 39 40 23 8. Sleep deprivation, sleep disorders, fatigue, stress and the risk of occupational injuries and errors 41 24 among health-care workers: a multi-approach epidemiological study. 42 25 9. Ghalichi L, Pournik O, M G. Sleep quality among health care workers. Arch Iran Med 43 44 26 2013;16(2):100-03. 10. Buxton OM, Marcelli E. Short and long sleep are positively associated with obesity, diabetes,

45 27 on September 27, 2021 by guest. Protected copyright. 46 28 hypertension, and cardiovascular disease among adults in the United States. Social Science & 47 48 29 Medicine 2010;71(5):1027-36. 49 30 11. Irwin MR. Why sleep is important for health: a psychoneuroimmunology perspective. Annu Rev 50 31 Psychol 2015;66:143-72. doi: 10.1146/annurev-psych-010213-115205 [published Online 51 32 First: 2014/07/26] 52 53 33 12. Gottlieb DJ, Redline S, Nieto FJ, et al. Association of usual sleep duration with hypertension: the 54 34 Sleep Heart Health Study. Sleep 2006;29(8):1009-14. 55 35 13. Kling RN, McLeod CB, Koehoorn M. Sleep problems and workplace injuries in Canada. Sleep 56 57 36 2010;33(5):611-8. doi: 10.1093/sleep/33.5.611 [published Online First: 2010/05/18] 58 37 14. Eddy R. Sleep deprivation among physicians. British Columbia Medical Journal 2005(4):176-80. 59 38 15. Ayas NT, Barger LK, Cade BE, et al. Extended work duration and the risk of self-reported 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 42

1 2 3 1 percutaneous injuries in interns. JAMA 2006;296(9):1055-62. doi: 10.1001/jama.296.9.1055 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 [published Online First: 2006/09/07] 6 3 16. Chaudhury PK, Warnock GL, Whalen TV, et al. CAGS and ACS evidence based reviews in surgery. 7 4 Risks of complications by attending physicians after performing nighttime procedures. Can J 8 9 5 Surg 2012;55(5):337-9. doi: 10.1503/cjs.018212 [published Online First: 2012/09/21] 10 6 17. Mycyk MB, McDaniel MR, Fotis MA, et al. Hospitalwide adverse drug events before and after 11 7 limiting weekly work hours of medical residents to 80. Am J Health Syst Pharm 12 8 2005;62(15):1592-5. doi: 10.2146/ajhp040527 [published Online First: 2005/07/21] 13 14 9 18. Lockley SW, Landrigan CP, Barger LK, et al. When policy meets physiology: the challenge of 15 10 reducing resident work hours. Clin Orthop Relat Res 2006;449:116-27. doi: 16 11 10.1097/01.blo.0000224057.32367.84 [published Online First: 2006/06/14] 17 18 12 19. Department ofFor Economic andpeer Social Affairs review Population Division. only World population ageing 2015. 19 13 United Nations. 2015; 135-154.18. Chen W, Zheng R, Baade PD, et al. 20 14 20. Cancer statistics in China, 2015. CA Cancer J Clin 2016;66(2):115-32. doi: 10.3322/caac.21338 21 22 15 [published Online First: 2016/01/26]20. Chen W, Zheng R, Baade PD, et al. Cancer statistics in 23 16 China, 2015. CA Cancer J Clin 2016;66(2):115-32. doi: 10.3322/caac.21338 [published Online 24 17 First: 2016/01/26] 25 18 21. Moran A, Gu D, Zhao D, et al. Future cardiovascular disease in china: markov model and risk factor 26 27 19 scenario projections from the coronary heart disease policy model-china. Circ Cardiovasc 28 20 Qual Outcomes 2010;3(3):243-52. doi: 10.1161/CIRCOUTCOMES.109.910711 [published 29 21 Online First: 2010/05/06] 30 31 22 22. Kobayashi T, Suzuki E, Takao S, et al. Long working hours and metabolic syndrome among 32 23 Japanese men: a cross-sectional study. Bmc Public Health 2012;12(1):395. 33 24 23. Kim MS, Kim T, Lee D, et al. Mental disorders among workers in the healthcare industry: 2014 34 35 25 national health insurance data. Ann Occup Environ Med 2018;30:31. doi: 10.1186/s40557- 36 26 018-0244-x [published Online First: 2018/05/15]

37 27 24. Andersen LL, Clausen T, Carneiro IG, et al. Spreading of chronic pain between body regions: http://bmjopen.bmj.com/ 38 28 prospective cohort study among health care workers. Eur J Pain 2012;16(10):1437-43. doi: 39 40 29 10.1002/j.1532-2149.2012.00143.x [published Online First: 2012/03/31] 41 30 25. Jank R, Gallee A, Boeckle M, et al. Chronic Pain and Sleep Disorders in Primary Care. Pain Res Treat 42 31 2017;2017:9081802. doi: 10.1155/2017/9081802 [published Online First: 2018/02/08] 43 44 32 26. Stonge MP, Mikic A, Pietrolungo CE. Effects of Diet on Sleep Quality. Advances in Nutrition 2016;7(5):938.

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 27. Buysse DJ. Sleep health: can we define it? Does it matter? Sleep 2014;37(1):9. 47 48 35 28. Knutson KL, Phelan J, Paskow MJ, et al. The National Sleep Foundation's Sleep Health Index. Sleep 49 36 Health 2017;3(4):234. 50 37 29. Consensus Conference P, Watson NF, Badr MS, et al. Joint Consensus Statement of the American 51 38 Academy of Sleep Medicine and Sleep Research Society on the Recommended Amount of 52 53 39 Sleep for a Healthy Adult: Methodology and Discussion. Sleep 2015;38(8):1161-83. doi: 54 40 10.5665/sleep.4886 [published Online First: 2015/07/22] 55 41 30. Seyedmehdi SM, Rahnama N, Yazdanparast T, et al. Prevalence of snoring and the risk of sleep 56 57 42 apnea in hospital staff. Work 2016;55(4):765-72. doi: 10.3233/WOR-162460 [published 58 43 Online First: 2017/01/07] 59 44 31. National Bureau of Statistics. China health statistical yearbook 2017. Beijing Union Medical 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 42 BMJ Open

1 2 3 1 University Press. 2017; 118:138. BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 32. Chinese Medical Doctor Association. Chinese doctors practicing white paper 2015. 6 3 http://www.cmda.net/zlwqgzdt/596.jhtml [Accessed 14 Apr 2018]. 7 4 33. Jafari Roodbandi A, Choobineh A, Daneshvar S. Relationship between circadian rhythm amplitude 8 9 5 and stability with sleep quality and sleepiness among shift nurses and health care workers. 10 6 Int J Occup Saf Ergon 2015;21(3):312-7. doi: 10.1080/10803548.2015.1081770 [published 11 7 Online First: 2015/09/29] 12 8 34. ALEMOHAMMAD ZB, TAGHAVI SME, SHARIFIAN A. Sleep Quality and Restless Legs Syndrome 13 14 9 among Health-care Workers: Shift Workers and Non-shift Workers. Journal of Sleep Sciences 15 10 2016;1(2):74-79. 16 11 35. Kolo ES, Ahmed AO, Hamisu A, et al. Sleep health of healthcare workers in Kano, Nigeria. Niger J 17 18 12 Clin Pract For2017;20(4):479-83. peer doi: 10.4103/1119-3077.204378review only [published Online First: 19 13 2017/04/14] 20 14 36. National Health and Family Planning Commission. China health statistical yearbook 2015. Beijing: 21 22 15 China Union Medical University Press, 2015; 11. 23 16 37. Stranges S, Tigbe W, Gomez-Olive FX, et al. Sleep problems: an emerging global epidemic? 24 17 Findings from the INDEPTH WHO-SAGE study among more than 40,000 older adults from 8 25 18 countries across Africa and Asia. Sleep 2012;35(8):1173-81. doi: 10.5665/sleep.2012 26 27 19 [published Online First: 2012/08/02] 28 20 38. Merskey H. International Association for the Study of Pain: Classification of chronic pain. 29 21 Descriptions of chronic pain syndromes and definitions of pain terms. Pain 1986; 3(2):S1- 30 31 22 S226. 32 23 39. Chen W, Zhou F, Hall BJ, et al. Is there a relationship between geographic distance and uptake of 33 24 HIV testing services? A representative population-based study of Chinese adults in 34 35 25 Guangzhou, China. PLoS One 2017;12(7):e0180801. doi: 10.1371/journal.pone.0180801 36 26 [published Online First: 2017/07/21]

37 27 40. Wang J, Xie H, Fisher JH. Multilevel models: applications using SAS2009. http://bmjopen.bmj.com/ 38 28 41. Ju H, Pan XL, Li A. Multi-level modelling of the factors that influence the participation of disabled 39 40 29 rural individuals in social medical insurance in China. Bmc Health Services Research 41 30 2013;13(1):58-58. 42 31 42. Ertel KA, Berkman LF, Buxton OM. Socioeconomic status, occupational characteristics, and sleep 43 44 32 duration in African/Caribbean immigrants and US White health care workers. Sleep 2011;34(4):509-18. [published Online First: 2011/04/05]

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 43. Burns RD. Energy balance-related factors associating with adolescent weight loss intent: evidence 47 48 35 from the 2017 National Youth Risk Behavior Survey. BMC Public Health 2019;19(1):1206. doi: 49 36 10.1186/s12889-019-7565-8 [published Online First: 2019/09/04] 50 37 44. Onwuekwe I, Onyeka T, Aguwa E, et al. Headache prevalence and its characterization amongst 51 38 hospital workers in Enugu, South East Nigeria. Head & face medicine 2014;10(1):48. 52 53 39 45. Holland PR. Headache and sleep: shared pathophysiological mechanisms. Cephalalgia 54 40 2014;34(10):725-44. doi: 10.1177/0333102414541687 [published Online First: 2014/07/24] 55 41 46. Zhang Y, Punnett L, Mcenany GP, et al. Contributing influences of work environment on sleep 56 57 42 quantity and quality of nursing assistants in long-term care facilities: A cross-sectional study. 58 43 Geriatric Nursing 2015;37(1):13-18. 59 44 47. Burgess HJ, Sharkey KM, Eastman CI. Bright light, dark and melatonin can promote circadian 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 42

1 2 3 1 adaptation in night shift workers. Sleep Medicine Reviews 2002;6(5):407-20. BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 5 2 48. Neil-Sztramko SE, Pahwa M, Demers PA, et al. Health-related interventions among night shift 6 3 workers: a critical review of the literature. Scandinavian Journal of Work Environment & 7 4 Health 2014;40(6):543-56. 8 9 5 49. Li J, Bidlingmaier M, Petru R, et al. Impact of shift work on the diurnal cortisol rhythm: a one-year 10 6 longitudinal study in junior physicians. J Occup Med Toxicol 2018;13:23. doi: 10.1186/s12995- 11 7 018-0204-y [published Online First: 2018/08/21] 12 8 50. Rebecca SC, Howard SK, Mac DT, et al. Improving alertness and performance in emergency 13 14 9 department physicians and nurses: the use of planned naps. Annals of Emergency Medicine 15 10 2006;48(5):596-604.e3. 16 11 51. Basner M, Fomberstein KM, Razavi FM, et al. American Time Use Survey: Sleep Time and Its 17 18 12 RelationshipFor to Waking peer Activities. Sleepreview 2007;30(9):1085-95. only 19 13 52. Chinese Medical Doctor Association. Chinese doctors practicing white paper 2017. 20 14 http://www.cmda.net/u/cms/www/201807/06181247ffex.pdf. [Accessed 20 May 2018] 21 22 15 53. Hale L. Who has time to sleep? J Public Health 2005;27(2):205-11. 23 16 54. Lancet T. Doctors’ training and the European Working Time Directive. Lancet 24 17 2010;375(9732):2121. 25 18 55. Ren Y. Study on Labor Rights Guarantee of Medical Staff in China. Legal System and Society 26 27 19 2017(4):293-94. 28 20 56. Patel SR, Hu FB. Short sleep duration and weight gain: a systematic review. Obesity 29 21 2008;16(3):643. 30 31 22 57. Tan X. Effects of exercise and diet interventions on obesity-related sleep disorders in men: study 32 23 protocol for a randomized controlled trial. Trials 2013;14(1):1-9. 33 24 58. Han Y, Yuan Y, Zhang L, et al. Sleep disorder status of nurses in general hospitals and its 34 35 25 influencing factors. Psychiatria Danubina 2016;28(2):176-83. 36 26 59. Organization WH. Global recommendations on physical activity for health. 2010

37 27 60. Shan HP, Yang XH, Zhan XL, et al. Overwork is a silent killer of Chinese doctors: a review of Karoshi http://bmjopen.bmj.com/ 38 28 in China 2013-2015. Public Health 2017;147:98. 39 40 29 61. Thomas SJ, Calhoun D. Sleep, insomnia, and hypertension: current findings and future directions. J 41 30 Am Soc Hypertens 2017;11(2):122-29. doi: 10.1016/j.jash.2016.11.008 [published Online 42 31 First: 2017/01/23] 43 44 32 62. Yu H, Xie LF, Chen K, et al. Initiating Characteristics of Early-onset Type 2 Diabetes Mellitus in Chinese Patients. Chin Med J (Engl) 2016;129(7):778-84. doi: 10.4103/0366-6999.178959

45 33 on September 27, 2021 by guest. Protected copyright. 46 34 [published Online First: 2016/03/22] 47 48 35 63. Bethany AB JD. Alternatives for Analysis of Complex Sample Surveys: A Comparison of SAS, 49 36 SUDAAN and AM software. SAS Global Forum 2007:2659-66. 50 51 37 52 53 54 38 Figure legend 55 56 39 Figure 1. Subjective sleep duration (hours per night) 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 42 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 Figure 1. Subjective sleep duration (hours per night) 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 27, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 42 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 Additional file 1. Questionnaire 5 6 7 8 ID number：______9 10 [1]Name：______11 ： 12 [2]Age ______13 [3]Gender：_____ ①Male ②Female 14 15 [4]Hospital：______16 For peer review only 17 [5]Department： 18 19 [5.1]Clinical department：______(If you are not a healthcare worker in the clinical 20 21 department, please fill in 0.) 22 23 ①Internal Medicine ②Surgery Department 24 25 ③Obstetricians and Gynecologists ④Pediatrics Department 26 27 ⑤Ophthalmology Department ⑥Stomatology Department 28 29 ⑦Otolaryngology Department ⑧Dermatology Department 30 31 ⑨Anesthesiology Department ⑩Emergency Department 32 ⑪Traditional Chinese Medicine Department ⑫Infectious Diseases Department 33 http://bmjopen.bmj.com/ 34 35 ⑬Plastic Surgery Department ⑭ICU ⑮Others ______36 [5.2] Medical technology department：______(If you are not a healthcare worker in 37 38 the medical technology department, please fill in 0.) 39 40 ①Pharmacy Department ②Clinical Lab ③Radiology Department 41 on September 27, 2021 by guest. Protected copyright. 42 43 ④Pathology Department ⑤Nuclear Medicine Department ⑥Nutritional Department 44 45 ⑦Others______46 47 [5.3] Administrative department：______(If you are not a healthcare worker in the 48 49 medical technology department, please fill in 0.) 50 51 ①Personnel Department ②Preventive care department 52 53 ③Medical Service Management Office ④Finance Department 54 55 ⑤Administration Office ⑥Others______56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 42 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 [6]Educational background：______5 6 ①Doctor ②Master ③Bachelor ④Junior college 7 8 ⑤High school or technical secondary school ⑥Others______9 10 [7]Title：______11 ① ② ③ ④ 12 senior title intermediate job title junior titles No title 13 ⑤Others______14 15 [8] What is your monthly salary now? ______yuan 16 For peer review only 17 [9] How much time do you work on average per week? ______Hours 18 19 [10] What's your average monthly night shift (over half an hour and above)？ 20 21 ______time(s)/per month 22 23 [11] How much time do you spend on your daily one-way commute? ______24 25 ①<30 minutes ②30-59 minutes ③60-89 minutes ④>=90 minutes 26 27 [12] Was your diet regular at work? ______28 29 ①Yes ②No 30 31 [13] Did you defecate regularly at work? ______32 ①Yes ②No 33 http://bmjopen.bmj.com/ 34 35 [14] Do you have time for physical exercise every day? ______36 ①Yes ②No 37 38 [15] How many hours do you usually sleep a day in the past month？______hour(s) 39 40 [16] Overall, in the last 30 days, how much of a problem did you have with sleeping,

41 on September 27, 2021 by guest. Protected copyright. 42 such as falling asleep, waking up frequently during the night, or waking up too early in 43 44 the morning? ______45 46 ①None ②Mild ③Moderate ④Severe ⑤Extreme 47 48 [17]Do you have experienced chronic pain that had not improved in the past 30 days in 49 50 any of the following regions？ 51 52 [Note: the chronic pain was defined as any continuous or persistent intermittent pain 53 54 experienced for a period longer than 3 months] 55 56 [17.1] Headaches ______①Yes ②No 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 42 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

1 2 3 4 [17.2] Neck and shoulder pain ______①Yes ②No 5 6 [17.3] Lumbar and back pain ______①Yes ②No 7 8 [17.4] Knee pain ______①Yes ②No 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 27, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 Additional file 2. Individual-level variables and coding 5 6 7 8 9 Table S1. Individual-level variables and coding for regression modeling. 10 11 Variables Characteristic Coding 12 Sex 13 Women 0 14 15 Men 1 16 Age (years) 17 < 30 1 18 For peer review30–40 only 2 19 20 > 40 3 21 Educational background 22 Bachelor degree and below 0 23 24 Master's degree and above 1 25 Job title 26 Junior titles and others 1 27 28 Intermediate title 2 29 Senior title 3 30 Monthly salary 31 ≤ 3,000 yuan 0 32 33 > 3,000 yuan 1 34 One-way commute time 35 < 30 minutes 0 36 ≥ 30 minutes 1

37 http://bmjopen.bmj.com/ 38 Regular diet at work 39 No 0 40 41 Yes 1 42 Work hours per week 43 (hours) 44 < 55 0

45 on September 27, 2021 by guest. Protected copyright. 46 ≥ 55 1 47 Night shifts per month 48 (times) 49 50 > 5 0 51 ≤ 5 1 52 Daily exercise 53 54 No 0 55 Yes 1 56 Headaches 57 No 0 58 59 Yes 1 60 Neck and shoulder pain

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 42

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 No 0 5 Yes 1 6 Lumbar and back pain 7 No 0 8 9 Yes 1 10 Knee pain 11 No 0 12 13 Yes 1 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 42 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from 4 Additional file 3. The names of Ethics Committees 5 6 7 Coding The names of Ethics Committees 8 9 1 The Ethics Committee of Daqing Longnan Hospital 10 2 The Ethics Committee of General Hospital of Daqing Oil Field 11 12 3 The Ethics Committee of Daqing City People's Hospital 13 4 The Ethics Committee of Harbin 242 Hospital 14 15 5 The Ethics Committee of The Fifth Hospital of Harbin 16 6 The Ethics Committee of Harbin The First Hospital 17 7 The Ethics Committee of Harbin First Specialist Hospital 18 For peer review only 19 8 The Ethics Committee of Harbin Red Cross Central Hospital 20 9 The Ethics Committee of The Third Affiliated Hospital of Harbin Medical University 21 22 10 The Ethics Committee of The Fourth Affiliated Hospital of Harbin Medical University 23 11 The Ethics Committee of The First Affiliated Hospital of Harbin Medical University 24 25 12 The Ethics Committee of Heilongjiang Land Reclamation Bureau Hongxinglong Bureau Central Hospital 26 13 The Ethics Committee of The Fourth Hospital of Heilongjiang Province 27 28 14 The Ethics Committee of Heilongjiang Province Land Reclamation Headquarters General Hospital 29 15 The Ethics Committee of Heilongjiang Red Cross Sengong General Hospital 30 31 16 The Ethics Committee of Heilongjiang Provincial Hospital of Traditional Chinese medicine 32 17 The Ethics Committee of The Second Affiliated Hospital of Heilongjiang University of Chinese Medicine 33 34 18 The Ethics Committee of The First Affiliated Hospital of Heilongjiang University of Chinese Medicine 35 19 The Ethics Committee of Jixi Mining Group General Hospital 36 37 20 The Ethics Committee of Jixi People's Hospital http://bmjopen.bmj.com/ 38 21 The Ethics Committee of Jixi Hospital of Traditional Chinese Medicine 39 40 22 The Ethics Committee of The Second Affiliated Hospital of Jiamusi University 41 23 The Ethics Committee of Jiamusi City Tuberculosis Hospital 42 43 24 The Ethics Committee of Jiamusi Center Hospital 44 25 The Ethics Committee of Hongqi Hospital Affiliated to Mudanjiang Medical University 45 on September 27, 2021 by guest. Protected copyright. 46 26 The Ethics Committee of Qiqihar Third Hospital 47 27 The Ethics Committee of the Second Affiliated Hospital of Qiqihar Medical College 48 49 28 The Ethics Committee of Qiqihar Traditional Chinese Medicine Hospital 50 29 The Ethics Committee of the Third Affiliated Hospital of Qiqihar Medical University 51 52 30 The Ethics Committee of Heilongjiang Provincial Hospital 53 31 The Ethics Committee of Shuangyashan Coal General Hospital 54 55 32 The Ethics Committee of The Shuangyashan Mining Bureau General Hospital 56 33 The Ethics Committee of Yichun Forestry Central Hospital 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open Page 40 of 42

1 2 STROBE 2007 (v4) Statement—Checklist of items that should be included in reports of cross-sectional studies 3 4 5 Item Section/Topic Recommendation Reported on page # 6 # 7 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract Page 4, line 11 8 9 (b) Provide in the abstract an informative and balanced summary of what was done and what was found Page 4, line 18-21; 10 Page 4, line 22; 11 Page 5, line 1- 5. 12 For peer review only 13 Introduction 14 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported Page 7-9; 15 Page 10 line 1 16 17 Objectives 3 State specific objectives, including any prespecified hypotheses http://bmjopen.bmj.com/ Page 10, line 2-12 18 Methods 19 20 Study design 4 Present key elements of study design early in the paper Page 10, line 14-23; 21 Page 9, line 1-19; 22 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data Page 10, line 14-23; 23 collection Page 11, line 1-8 24 25 Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of participants on September 27, 2021 by guest. Protected copyright. Page 11, line 4-8 26 27 28 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if Page 11, line 10-22; 29 applicable Page 12; 30 Page 13, line 2-10 31 32 Page 14, line 6-10 33 Data sources/ 8* For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe Page 11, line 10-22; 34 measurement comparability of assessment methods if there is more than one group Page 12; 35 Page 13, line 2-10 36 37 Page 14, line 6-10 38 Bias 9 Describe any efforts to address potential sources of bias No 39 Study size 10 Explain how the study size was arrived at No 40 41 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and Page 11, line 10-22; 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

Page 41 of 42 BMJ Open

1 2 why Page 12; 3 Page 13, line 2-10 4 5 Page 14, line 6-10 6 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding Page 13, line 12-22; 7 Page 14, line 1-14; 8 (b) Describe any methods used to examine subgroups and interactions No 9 10 (c) Explain how missing data were addressed Page 11, line 6-8 11 (d) If applicable, describe analytical methods taking account of sampling strategy Page 10, line 21-23; 12 For peer review only Page 11, line 1-8 13 14 (e) Describe any sensitivity analyses No 15 Results 16 17 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, examinedhttp://bmjopen.bmj.com/ for eligibility, Page 10, line 21-23; 18 confirmed eligible, included in the study, completing follow-up, and analysed Page 11, line 1-8 19 (b) Give reasons for non-participation at each stage No 20 (c) Consider use of a flow diagram No 21 22 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential Page 16, line 9-15 23 confounders 24 (b) Indicate number of participants with missing data for each variable of interest No 25 on September 27, 2021 by guest. Protected copyright. Page 16, line 9-15; 26 Outcome data 15* Report numbers of outcome events or summary measures 27 Page 17, line 2-5 28 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence Page 17, line 7-22; 29 interval). Make clear which confounders were adjusted for and why they were included Page 18, line 1-9 30 31 (b) Report category boundaries when continuous variables were categorized Page 11, line 10-22; 32 Page 12; 33 Page 13, line 2-10 34 Page 14, line 6-10 35 36 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period No 37 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses Page 19, line 3-8; 38 Page 21, line 3-6 39 40 Discussion 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2019-032239 on 29 December 2019. Downloaded from

BMJ Open Page 42 of 42

1 2 Key results 18 Summarise key results with reference to study objectives Page 22, line 3-8 3 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and Page 26, line 15-22; 4 5 magnitude of any potential bias Page 27, line 1-6 6 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from Page 22-25 7 similar studies, and other relevant evidence Page 26, line 15-22; 8 Page 27, line 1-6 9 10 Generalisability 21 Discuss the generalisability (external validity) of the study results Page 26, line 15-22; 11 Page 27, line 1-6 12 Other information For peer review only 13 14 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on Page 28, line 6 15 which the present article is based 16 17 http://bmjopen.bmj.com/ 18 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and unexposed groups in cohort and cross-sectional studies. 19 20 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE 21 checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 22 23 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org. 24 25 on September 27, 2021 by guest. Protected copyright. 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60